(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Biodiversity Heritage Library | Children's Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "Wages, prices, profits, and productivity"

.... . :... 



Wages 

Prices 

Profits 

and 

Productivity 







The American Assembly 
Columbia University 






UNIVERSITY 
OF FLORIDA 
LIBRARIES 




This copy is distributed with the compliments of 

The Calvin K. Kazanjian Economics Foundation, Inc., 

Westport, Connecticut. 



Digitized by the Internet Archive 
in 2013 



http://archive.org/details/wagespricesprofiOOamer 



Wages 

Prices 

Profits 

and 

Productivity 



^^rf**"***™*"* 




The American Assembly 
Columbia University 
June 1959 



Final Edition 

Background papers 
and the Final Report 
of the Fifteenth 
American Assembly, 
Arden House, 
Harriman Campus of 
Columbia University, 
Harriman, New York 
April 30-May 3, 1959 



3 2 r, os' a. 



PRINTED IN U.S.A. 

Library of Congress Catalog Card No. 59-12574 
Copyright, 1959, The American Assembly 



Preface 



Wages, Prices, Profits and Productivity was the subject of the Fifteenth 
American Assembly at Arden House, Harriman, New York, April 30- 
May 3, 1959. About sixty Americans from business, labor, government, 
agriculture, education, communications and the clerical, legal and military 
professions engaged in three days of round table discussions, ending with 
a final report approved in plenary session on the fourth day. That report 
appears in this volume beginning on page 181. 

Formal addresses were given at the Assembly by Secretary of Labor 
James P. Mitchell, and Sumner Slichter of Harvard University. 

As with previous American Assemblies, so with the Fifteenth: the 
national meeting at Arden House will be followed by at least three regional 
assemblies. 

The papers which follow, prepared under the editorial supervision of 
Dr. Charles Myers of Massachusetts Institute of Technology, served as 
background for the Fifteenth Assembly and will be used at the regional 
meetings. The essays reflect the thought and experience of the individual 
authors. The American Assembly itself takes no official stand on the 
subjects of its sessions. 

Basic research amounting to one-tenth of the cost of the Fifteenth 
Assembly program was paid for by a grant from the Rockefeller Foun- 
dation. Other costs are to be met by a grant from the Maurice and Laura 
Falk Foundation of Pittsburgh. The Assembly expresses its thanks and 
warm appreciation. 

The American Assembly 
Henry M. Wriston 
President 



Contents 



Introduction: 

Central issues in wage-price relationships 1 

Charles A. Myers, Editor 

1 . Patterns of wages, prices and productivity 1 1 
Albert Rees 

LEARNING FROM HISTORY 11 
THE HISTORICAL RECORD: EARNINGS 12 
THE HISTORICAL RECORD: PRODUCTIVITY 20 
INTERPRETING THE RECORD 27 
CONCLUSIONS 34 

2. Productivity, costs and prices: 37 

Concepts and measures 
John W. Kendrick 

PRODUCTIVITY CONCEPTS 38 

CONCEPTS OF FACTOR COST OR INCOME 46 

CONCEPT OF THE PRICE LEVEL 49 

SOURCES AND METHODS OF MEASUREMENT 50 

NEED FOR STATISTICAL IMPROVEMENTS 56 

vii 



3. Underlying factors in the postwar inflation 61 

James S. Duesenberry 

INTRODUCTION 61 

THEORIES OF INFLATION 64 

PRICE MOVEMENT SINCE 1950 67 

THE CAUSES OF WAGE INFLATION 78 

THE SOURCES OF DEMAND 84 

CAN WE CONTROL INFLATION BY CONTROLLING DEMAND? 87 

CONCLUSION 89 

4. The impacts of unions on the level of wages 91 

Clark Kerr 

TYPES, CIRCUMSTANCES AND IMPACTS 94 
THE VARIETY OF EXPERIENCE 99 
OBSERVATIONS 105 
REMEDIES 107 



5. Wage behavior and inflation: 109 

An international view 

Lloyd G. Reynolds 

THE MECHANISM OF COST INFLATION 113 
STATISTICAL INDICATORS OF COST INFLATION 122 
STRUCTURAL DETERMINANTS OF COST INFLATION 125 
WHAT TO DO ABOUT IT? 131 

viii 



6. Policy Problems: 1 37 

Choices and proposals 
John T. Dunlop 

TYPES OF POLICY PRESCRIPTIONS 138 
THE COMMUNITY'S PRIORITIES AND PREFERENCES 140 
THE WAGE BARGAINING INSTITUTIONS 144 
STRATEGIC SECTORS OF SECULAR INFLATION 149 
SUMMARY 159 



Address to the nation from Arden House: 

Productivity and the Consumer 1 61 

James P. Mitchell 



Address to Fifteenth American Assembly: 

Labor costs and prices 1 67 

Sumner H. Slichter 



Final report of the Fifteenth American Assembly 181 
Participants in the Fifteenth American Assembly 187 
The American Assembly 191 



IX 



Introduction: 



Central issues in wage-price 
relationships 



Charles A. Myers Editor 



Are prices being pushed up by wage increases, or are wages simply 
following prices upward in the spiral of creeping inflation caused by other 
factors? These are simple and controversial questions which mask a com- 
plexity of factual data and analysis. The facts — and the blame — are hotly 
debated in the public press, by labor and management at collective bar- 
gaining sessions and also publicly, and even by professional economists 
who do not always agree among themselves. 

Furthermore, the issues involved in the controversy are important for 
public policy. Widespread concern for the maintenance of "a stable dollar" 



Charles A. Myers is Professor of Industrial Relations and Director of the Industrial 
Relations Section in the Department of Economics and Social Science, Massachusetts 
Institute of Technology. 

He has been at M.I.T. since 1939, except for absences during the War when he 
served in the Labor Division of the War Production Board and subsequently as a 
Public Panel Member of the War Labor Board. He was a Public Member of the 
Regional Wage Stabilization Board in 1952-53, and has served as arbitrator in 
numerous labor disputes since 1944. Between 1949 and 1956 he was a member 
of the Committee on Labor Market Research of the Social Science Research Council 
and served as an elected member of the Executive Board of the Industrial Relations 
Research Association in 1952-55. 

He is author or co-author of the following: Movement of Factory Workers 
(1943), The Dynamics of a Labor Market (1951), Industrial Relations in Sweden 
(1951), Personnel Administration (1956 3rd ed.), Labor Problems in the Industrial- 
ization of India (1958), as well as many other articles on wages, labor mobility, 
and labor-management relations. 



and the avoidance of further inflation has emphasized not only the impor- 
tance of government's monetary and fiscal policies (as the Fourteenth 
American Assembly considered), but also the need for examination of the 
relationships between wages, productivity, profits, and prices. In his State 
of the Union message to the Eighty-sixth Congress, President Eisenhower 
called attention to the importance of a balanced federal budget to maintain 
price stability and urged labor and management to exercise restraint in 
making wage and price increases in order "to curb the wage-price spiral." 
The President has also appointed a Cabinet Committee on Price Stability 
for Economic Growth, headed by Vice-President Nixon, and there are 
several other governmental committees working on various aspects of this 
problem. Congressional committees continue to hold hearings, so that 
wage-price issues are kept hot on the front burner. And if all this were not 
enough, collective bargaining negotiations in 1959, particularly in steel, 
promise front-page news and editorial comment on these issues. 

We need to examine carefully, therefore, the facts on wages, prices, 
profits and productivity, to assess their interrelationships, and to try to 
unravel some of the causal connections among them. Those involved 
directly in the wage and price-setting mechanisms seem to find little diffi- 
culty in pointing to a particular "culprit." But sober examination of the 
facts and their patient analysis reveal much more complexity and suggest 
the need for public policies which are both realistic and workable, even 
though they will probably not be panaceas for the inflationary pressures 
confronting the American economy. 

It is the purpose of the papers presented in this volume to examine both 
the facts and their significance for policy. Each of the authors has devoted 
a considerable portion of his professional life to the study of wage-price 
problems. Several have participated directiy in earlier governmental efforts 
to stabilize wages and prices in periods of national emergency. Further- 
more, some of them have studied at first-hand the experience of other 
countries in attempting to maintain stable prices through voluntary restraint 
or direct control of wage and price increases. Their combined study and 
judgment on these controversial issues provide the basis for a more in- 
formed discussion of appropriate public policies. 

The problem of measurement 

How much has productivity increased per year, and over different 
periods? What has been the increase in the "cost of living," in average 
hourly money earnings, in "real wages"? To answer these questions with 
reasonable accuracy, we must consider problems of definition and measure- 
ment. One of the very great difficulties is that most measures have a 
number of limitations, yet these are often lost sight of in the battleground 
of words about what has happened or will happen. 



"Labor productivity" is used loosely in many discussions, but until 
recently the most common measurement was "output per man-hour" of 
production workers, published by the Bureau of Labor Statistics of the 
United States Department of Labor. One of the limitations of this measure- 
ment is that the proportion of non-production employees, such as white- 
collar workers, research and development staffs, etc., has been increasing. 
Consequendy, the proportion of production workers to total output has 
been declining, and the index of output per man-hour of production work- 
ers has an upward bias; it does not accurately reflect the total employee 
productivity. Kendrick's paper brings out this point as well as many other 
difficulties in measuring productivity. Output per man-hour, he empha- 
sizes, is only a partial productivity measurement, since it is necessary to 
include inputs of capital to get a measurement of real product per unit 
of total factor input. This will measure changes in productivity efficiency 
generally. 

The BLS has recently released estimates of productivity which measure 
real product per employee man-hour paid for the private economy. These 
differ from its earlier index of physical output per production worker man- 
hour. Even this measure, as both Kendrick and the BLS have emphasized, 
does not measure "labor efficiency" in the sense of greater labor skill or 
effort in increasing output. The index will move upward for a variety of 
reasons: (1) increases in plant output resulting from technological change 
and the increased amount of capital required; (2) improvements in mana- 
gerial as well as worker efficiency; (3) changes in methods, processes, and 
materials; (4) shifts from less efficient to more efficient plants and indus- 
tries; and (5) shifts in the relative importance of industries with different 
levels of productivity. 

/ While the productivity index may move upward because workers in a 
particular industry may work harder or have more responsibility and skill 
in their jobs, it can move also upward for the other reasons (principally 
the first) . It will do so even though physical labor effort and job responsi- 
bility may be lower as a consequence of improved technology, processes 
and methods. It is important to emphasize that the measurement of pro- 
ductivity involves no ethical connotation of what workers or employees 
generally "contribute" to output and hence what they "should" receive in 
wages or salaries. 

The limitations of existing productivity indexes have led some specialists 
to concentrate on the development of new and better productivity measure- 
ments. The recent work of the BLS has already been noted; the National 
Bureau of Economic Research has also been in the forefront of this effort. 
Kendrick's paper outlines some of the alternative concepts and measure- 
ments, as well as their limitations, and summarizes the available data for 
the period since 1889. Among other points he stresses the difference in the 



level of productivity when output per man-hour worked is used as com- 
pared with output per man-hour paid for. The latter more accurately 
reflects the fact that the increase in paid holidays and paid vacations should 
be included, especially if comparisons are made with total average hourly 
earnings (including such hours not worked but paid for) . But existing data 
do not always permit such comparisons over a period of time, so that some 
productivity measurements are subject to an upward bias. Conversely, they 
underestimate real productivity to the extent that quality improvements 
cannot be measured accurately and taken into account in constructing 
the indexes. 

Similar difficulties arise in measuring changes in "the cost of living" or, 
more accurately, the Consumer Price Index of the BLS. Improvements in 
quality are difficult to take into account, and may give the index an upward 
bias. On the other hand, if there has been quality deterioration, a down- 
ward bias results. Furthermore, changes in the "market basket" of goods 
and services, as a result of changes in family spending patterns, are re- 
flected periodically in changes in the composition of the index. Some 
changes in the weights assigned to different items in 1947-49 were made 
in 1953 and minor ones in 1955. As a consequence, the importance of 
commodities in the index has declined while the importance of services has 
increased. If the composite index has risen, therefore, is it because the 
"cost of living" has risen or because the level of living ("the cost of better 
living") has risen? If better living is thus involved in the construction of 
the index, then it has an upward bias. Probably this bias is small, for the 
BLS has tried to construct an index which measures price changes only, 
and not a rise in the standard of living. 

Computations of average hourly money earnings also involve debatable 
questions. Should they be based on hours worked, or hours paid for? To 
be comparable with productivity measurements using "hours paid for," the 
earnings index should be computed from total compensation (including 
paid holidays and vacations), divided by hours worked. But what about 
the other "wage supplements" which Rees's paper mentions? These include 
employer contributions for unemployment compensation and supplemen- 
tary unemployment benefits, old-age and survivors' insurance, and private 
pension plans. If these are added, the index of total compensation rises 
in recent years. Finally, if the Consumer Price Index is not wholly satis- 
factory, then indexes of real earnings or real compensation are also 
partly inaccurate. 

The need for better statistics and research 

All of this points to the clear need for better statistical data on basic 
changes in the economic quantities which directly affect price and wage 
decisions. These are vital in each bargaining session as well as at the 



national level when public policies for dealing with such problems as the 
alleged "wage-cost push" are being hammered out. As Dr. Ewan Clague, 
United States Commissioner of Labor Statistics, has said, "The statistical 
tools which economists must use to analyze economic developments are not 
good enough to do the job which is needed." Will responsible repre- 
sentatives of labor and management, and of the Congress, demand better 
statistics, or will charges and counter-charges continue to be hurled with 
the support of inadequate or questionable data? The papers in this volume 
point to some of these difficulties, and illustrate how carefully couched and 
qualified our conclusions must be on the basis of all presently available 
information. Clearly, more study and research are needed on these im- 
portant issues. 

Trends in productivity, prices, profits and wages 

With all the qualifications just mentioned, the available historical record 
is, nevertheless, instructive. In addition to the productivity indexes sum- 
marized in Kendrick's tables, Rees presents yearly data from 1889 through 
1957 on average hourly money earnings in manufacturing, wage supple- 
ments per hour and total compensation per hour at work, together with the 
Consumer Price Index, total real compensation (computed in 1957 dollars 
per hour at work), and two productivity indexes. Some of these data for 
the earlier period are new, developed by Rees in a study for the National 
Bureau of Economic Research. 

Average hourly earnings in manufacturing (including paid holidays and 
vacations) have increased nearly 14 times during the 69-year period, to an 
average of $2.08 in 1957. Other wage supplements, computed since 1929, 
add almost 16 cents an hour to this figure. Consumer prices have tripled 
since 1889, as measured by a linking of several indexes. Because money 
earnings rose so much faster than consumer prices, real average hourly 
earnings in manufacturing (total real compensation in 1957 dollars per 
hour at work) rose five times between 1889 and 1957. 

During part of this period, output per man-hour in manufacturing rose 
more rapidly than real hourly earnings, as Rees points out, but after 1929, 
real earnings moved ahead of the productivity index. Taking Kendrick's 
productivity index of real product per unit of labor and capital combined 
(total factor input in the private economy), we find that real average 
hourly earnings rose at a faster rate than this measure of productivity after 
1913, and much faster after 1929. During the same period there has been 
no upward trend in the return on capital per unit, because the total stock 
of capital has increased relative to labor. 

None of these trends, it should be emphasized, explain the causal rela- 
tions. The fact that real average hourly earnings outstripped total factor 



productivity after 1913 does not necessarily mean that wage increases 
made voluntarily by employers during the period or negotiated with unions 
were the primary cause of rising prices. Nothing in these data tells us what 
caused the greater increase in real average hourly earnings. Clearly, unions 
were not strong enough between 1913 and 1929 to exert an independent 
influence on general wage levels, whatever the merits of the argument 
that they have so influenced wage levels in a more recent period. 

We need to look further and deeper, as does Rees in the last part of his 
paper and as do the other writers in the papers which follow. The picture 
then becomes more cloudy. Through the haze of complex interrelationships 
in the real world where "other things don't remain the same," we see that 
answers to our questions are not so clear cut, that the facts do not all 
point in one direction, and that informed judgment and evaluation are the 
best we can expect in the present state of our statistics and our knowledge. 

An outline of the principal wage-price issues 

Our purpose in this volume is to raise some of the most pressing issues 
for public discussion, and to present the best available evidence oh them. 
The task of the introductory essay is to pose the questions and not to 
provide answers. Some answers are suggested by the authors of particular 
papers, and these authors do not always completely agree among them- 
selves. Even informed judgment and analysis do not result in identical 
conclusions in this range of questions any more than they do in most other 
controversial and complex fields. Perhaps those who have all the answers 
do not start with all the relevant questions, or consider all the available 
evidence dispassionately. The first step toward more informed public dis- 
cussion of the issues, therefore, is to ask the right series of questions. The 
following may contribute to this goal. 

1. The measurements: their usefulness and limitations 

Accurate and generally understood measurements of productivity, wages, 
prices, and profits are needed for the process of setting wages and prices 
in hundreds and thousands of centers of decision-making. Effective and 
equitable public policies designed to maintain price stability and economic 
growth depend upon reliable data on these economic dimensions. But, as 
we have noted and as the papers bring out more clearly, there are im- 
portant limitations in the existing indexes, and they should be used with 
more caution than they have been in many public discussions. How can 
the adequacy and accuracy of these important indexes be improved? 
Should additional funds be appropriated to improve the collection of data, 
the construction of the indexes, and the widespread public dissemination 



of their meaning? To what extent can we rely on existing measurements 
without further improvement in their coverage or quality? Which of the 
available indexes of productivity and wages (or earnings) should be used 
in evaluating private decisions and in guiding public policies? Does a re- 
view of the long-term relationships between these indexes throw any light 
on the present controversy over "cost-push" or "wage-induced" inflation? 

2. The role of wages and prices in the postwar inflation 

The growth of a strong labor movement in the United States after 1933, 
the postwar wave of strikes and substantial wage settlements, and the 
tendency of unions and workers to expect periodic wage increases have all 
combined to focus public attention on rising wages as a cause of the post- 
war inflation. Union leaders and others have countered by pointing the 
ringer at industrial price policies and profits as more basic explanations of 
rising prices. In most public discussions, some weight is also given to de- 
mand factors, particularly to government expenditures. But the tendency is 
to assess the blame rather than to analyze the causes of the postwar inflation. 

What are the facts on the movements of wages and prices in the different 
sectors of the economy since 1947? How do the rates of change compare? 
Is there any significant distinction between parts of the 12-year period from 
the beginning of 1947 through 1958? How do movements in labor pro- 
ductivity compare with wage movements? To what extent does the "de- 
mand pull" explanation help us to understand the wage and price behavior 
of this period as compared to the "cost push" or "wage induced" ex- 
planation? What has been the role of profit margins in relation to price 
changes during this period? Can we say anything about the causal rela- 
tionship? What significance has the condition of the postwar labor market 
in general and in particular occupations or industries had in explaining 
wage movements? Were government monetary and fiscal policies respon- 
sible for the "demand pull" on costs during this period, or any part of it, 
and to what extent? Considering all of the evidence, what is the most 
generally valid explanation of the postwar inflation in the United States? 

3. Labor movements and their impact on wages and prices 

Unions in the United States are strongly organized in certain manu- 
facturing industries such as steel and autos, as well as in mining and heavy 
construction; but union membership is a much smaller proportion of the 
labor force in cotton textiles and chemicals, for example, and in the service 
industries generally, such as banking, insurance, finance, and trade. More- 
over, wage bargaining is seldom on a national or centralized basis, even 
though the phenomenon of "pattern bargaining" has evolved in the postwar 



period of full employment. But the labor movements of other countries are 
often different from ours; there is more centralized wage bargaining in 
Sweden and in the Netherlands, for example. Deliberate policies of "wage 
restraint" have been followed by these labor movements in some postwar 
years, as a part of a governmental program (often a labor government) for 
economic recovery or stability. Even within the American labor movement 
there have been different approaches to wage-price relations. Some unions 
appear unconcerned about the impact of wage demands on prices; others 
relate or profess to relate wage bargaining to the prosperity of the firm or 
industry, and its ability to remain competitive or to avoid price increases. 
What is the relationship of different types or classifications of unions 
and labor movements to wage-price policies? What has been their impact 
on wage changes and price-level changes? Are there significant differences 
between the types of labor movements and the changes in wages and 
prices? Does the available evidence point to the superiority of any one 
type of labor movement over the others in helping to maintain price 
stability? What implications does this have for general economic policies 
directed toward avoiding inflation? 

4. Conditions leading to a cost-push or wage-induced inflation 

A cost-push inflation may develop because some important cost com- 
ponent, such as the price of imported raw materials, rises as a result of 
external influences which may be caused by a demand pull (as during the 
Korean War). This example illustrates the interrelationships between de- 
mand and costs, and the great difficulty in attributing all of the influence 
to one factor. The cost-push influence may also result, as Reynolds points 
out in his paper, when some strong interest group in the economy strives 
to improve its relative position, as in the case of the farm bloc, certain 
manufacturers and retailers' organizations, as well as some trade unions. 

Is a cost push possible in these cases only when demand conditions are 
favorable, or made permissive by monetary and fiscal policy? If the mone- 
tary authorities act vigorously enough, can't they damp down demand 
sufficiently to discourage the cost push from separate interest groups? Or 
are there certain built-in structural causes of the cost push, particularly in 
wages? Is it realistic to expect that average hourly money earnings will go 
up just as fast as the average national rise in output per man-hour? If 
labor productivity is increasing faster in some industries, and if wages 
there are increased more than the national average increase in productivity 
for the economy as a whole, are wages likely to rise considerably less in 
the industries with slower increases in productivity? How do worker ex- 
pectations of further wage increases, and the structure of unionism and 
collective bargaining, affect this possibility? What about the impact of 

8 



long-term agreements for future wage increases, related either to changes in 
the Consumer Price Index, or to expected increases in productivity (as in 
the case of the "annual improvement factor" in the automobile collective 
agreements), or to fixed automatic increases at specified future dates 
regardless of economic conditions at a later time? Have all these develop- 
ments made wage costs more rigid in an economic downturn, and less 
likely to lag behind price increases during the upturn? These questions are 
central to an evaluation of the significance of so-called "wage-induced 
inflation," but do they cover the whole problem? Are wage increases the 
basis for even greater price increases in the industries characterized by 
so-called "administered prices"? To what extent are industrial price 
policies the real initiating factors in the cost push? 

5. Alternative policy proposals for dealing with "wage-in- 
duced inflation" 

Public discussions of the postwar inflation and its causes have been filled 
with proposals for dealing with that part of the cost push which is attribu- 
table by the critics to wage pressures from unions. These range from 
greater reliance on monetary and fiscal policies to damp down demand (in- 
cluding the demand for labor) to suggestions that the power of unions be 
curbed by a variety of methods. Sometimes, policy proposals contain a 
mixture of both. 

What is the likelihood that monetary and fiscal policies can be effective 
in reducing demand enough to discourage continued wage and price 
increases? Do we know what rate of unemployment is sufficient to 
accomplish this objective? What are the specific proposals for legislative 
restriction of union power over wages, and to what extent are these likely 
to be effective in achieving the objective of discouraging "wage-induced 
inflation"? Will other forms of union regulation (such as those designed 
to prevent corruption or encourage greater democracy) have any impact 
on union wage demands, and if so, in what direction? Alternatively, would 
unions be more "responsible" in their wage demands if collective bargain- 
ing were centralized on an economy-wide basis or industry-wide basis? 
Or, would the power of trade unions be more effectively offset if employers 
combined on a regional or industry-wide basis for collective bargaining, 
and resisted wage demands with more vigor — including willingness to 
take strikes? Would the public be willing to pay the cost in labor strife 
and inconvenience? 

Should government impose wage controls, as it did during World War 
II and the Korean War? If this is done, will price and profit controls also 
be necessary? Finally, are governmental pleas for "restraint" and "respon- 
sibility" in wage and price decisions likely to have much effect on unions 

9 



and employers? Can some procedure be devised for bringing key union 
and management officials together periodically with responsible government 
officials for discussions of the impact of price and wage policies on price 
stability? What are the prospects that such a procedure would be success- 
ful? Are there alternative procedures for dealing with the cost push in 
key sectors of the economy? 

6. Conflicting national goals and policy choices 

In our national policy since the war we have emphasized full employ- 
ment. But the postwar inflation has increased our concern for price 
stability, and proposals have been made to write this into the Full Employ- 
ment Act of 1946 as a twin and equally important objective. More recent- 
ly, the growing threat of Soviet economic supremacy has emphasized the 
necessity for the United States to achieve a higher rate of economic growth. 

To what extent are these three goals fully compatible? Or are we as a 
nation forced to make some hard and unpleasant choices? If we want price 
stability above all, will this require some greater percentage of unemploy- 
ment than the March, 1959, 5.8 per cent rate (seasonally adjusted), which 
means nearly 4.4 million unemployed workers? Can we even countenance 
this much unemployment for very long? If productivity has been lagging 
behind wages recently, as some contend, is this partly a consequence of a 
slower rate of economic growth? Would economic expansion and a higher 
rate of investment mean another demand-pull inflation bringing wages and 
prices along on an upward course? How much do we value the American 
system of trade unions and collective bargaining? In short, what price are 
the American people willing to pay to achieve price stability, or full em- 
ployment, or economic growth, or free collective bargaining and industrial 
peace— or some measure of each? 



10 



1 




Patterns of wages, prices and 
productivity 



Albert Rees 



Learning from history 

In the period since World War II, there has been intense interest in the 
pattern of behavior of wages, prices, and productivity — interest with which 
is mixed more than a little anxiety. It will help us to see whether there is 
now special cause for concern if we examine the recent past against the 
background of the long-term record. Are wages rising more rapidly now 
than in earlier times? Has the relation of wages to productivity changed? 
If so, why? This paper will look into these questions to see whether the 
answers, in so far as we know them, shed light on our present position. 
We shall see that there has in fact been a marked acceleration in the 
rate of growth of both real wages and productivity, but that since 1929 
real wages have been growing faster than productivity. This pattern can 
be reasonably well explained in terms of fundamental economic forces — 



Albert Rees is Associate Professor of Economics at the University of Chicago and 
editor of the Journal of Political Economy. He has served on the staff of the Council 
of Economic Advisers and as research associate of the National Bureau of Economic 
Research. 

In preparing this paper Dr. Rees benefited from discussion with C. F. Christ, 
H. G. Lewis, and G. P. Shultz of the University of Chicago. s 

11 



chiefly changes in the supplies of labor and capital — though part of the 
pattern may also be due to institutional changes in wage determination. 

The historical record: earnings 
Money earnings 

The figures on money earnings presented here are confined to manu- 
facturing wage earners. This is one of the largest groups of workers in 
the economy, one that has been heavily affected by the growth of unionism 
in the past twenty-five years, and one of the very few groups for which we 
have reasonably consistent earnings figures covering a long period. Even 
for this group, the record has gaps and defects, so that it has been neces- 
sary to construct a new wage series for this paper. 1 

The first column of Table 1 shows average hourly earnings of manu- 
facturing wage earners annually since 1889. The same definition of manu- 
facturing is used throughout. Thus railroad repair shops, which were 
considered a part of manufacturing before 1937, are excluded from all 
the data. 

Earnings have been measured per hour at work, so that increases in paid 
holidays, paid vacations, or paid sick leave increase average hourly earn- 
ings. In this respect, the figures presented are unlike the familiar ones 
of the Bureau of Labor Statistics, which measures earnings per hour paid 
for. In the BXS series, increases in paid vacations, holidays, or sick leave 
do not raise hourly earnings. 

Other forms of time paid for but not worked are not accounted for in 
the series shown here. Thus the figures understate the rise in earnings per 
hour of actual work to the extent that there has been an increase in pay for 
such things as lunch periods, coffee breaks, wash-up time, call-in time, and 
jury duty. According to a survey by the Chamber of Commerce of the 
United States, such items amounted to 2.5 per cent of payroll for manu- 
facturing firms in 1957. However, the firms surveyed may have made 
more generous payments for such time than did all manufacturing firms. 

The series on earnings per hour at work is based largely on data from 
the Census of Manufactures. In 1956, hours at work as measured by the 
Census Bureau were 5.4 per cent lower than hours paid for as reported by 
the BLS. This by itself should cause earnings per hour at work to be 
5.4 per cent higher than earnings per hour paid for. In addition to this 
conceptual difference, however, there is a difference in the sample of 



1 For the period 1890-1914, the wage series is taken from my larger study "Real 
Wages in Manufacturing, 1890-1914," done with the assistance of Donald P. Jacobs 
for the National Bureau of Economic Research. I am grateful to the National Bureau 
for permission to draw on this study and other unpublished studies. 

12 



establishments covered. In recent years, the sampling difference and the 
difference in the hours concept roughly offset one another. In the 1930's, 
only the sampling difference is important because few manufacturing wage 
earners received paid holidays or paid vacations. The result of this is 
that the new series rises more than the BLS series between 1939 and 1956. 
From 1889 to 1957 the money earnings of manufacturing wage earners 
increased from 15.1 cents per hour at work to $2.08, almost a fourteen- 
fold increase. However, this is not the whole story. We must take account 
of employer payments for pensions and insurance, which were unknown 
for wage earners in 1889 and are almost universal today. 

Wage supplements 

The second column of Table 1 presents new estimates of wage supple- 
ments per hour of work, beginning with 1929. Estimates of wage supple- 
ments have previously been available for this whole period only for wage 
earners and salaried workers together, and these were not expressed per 
man-hour. Wage supplements consist of employer contributions to old age 
and survivors insurance, unemployment insurance, workmen's compensa- 
tion, and private pension, welfare, and insurance plans. The figures do not 
include other items sometimes considered fringe benefits, such as Christmas 
bonuses, subsidies to company cafeterias, or discounts on goods bought 
from the company. 

Including wage supplements along with wages in figuring worker com- 
pensation implicitly assumes that the benefits provided by both public and 
private insurance and welfare plans are worth what they cost. Of course, 
this will not seem true to each worker individually. There are always some 
workers, or groups of workers, who would prefer to have cash. But unless 
most workers feel that wage supplements are worth what they cost, it is 
hard to account for their rapid growth in both union and non-union plants. 
Many forms of insurance can be bought at less cost for groups of workers 
than for individuals, and are in this sense "worth" more than they cost. 

Later I shall divide wage supplements, together with wages, by the 
Consumer Price Index to put them in dollars of constant purchasing power. 
It may be objected that this is incorrect to the extent that pensions or other 
benefits are not received at the time the contributions are made, and thus 
their real value depends on the price level when they are received. This 
does present a problem, but one for which there is no perfect solution. 
Moreover, the same objection applies to adjusting money earnings by a 
price index, to the extent that individual workers use money earnings to 
add to their savings. 

Data on wage supplements are not available before 1929. The 1929 
figure was only 0.4 cents per hour at work, most of .which probably repre- 

13 



sented the cost of workmen's compensation. The amount in earlier years 
must have been smaller still, and the error caused by its omission is 
negligible. 

From 1929 to 1957, the estimated cost of wage supplements per hour at 
work rose from 0.4 cents to 16.2 cents. The first big jump comes in the 
late 1930's, following the enactment of the Social Security law. After 
1943, private pensions, insurance, and welfare plans became increasingly 
important. 

Total compensation 

By adding money earnings and wage supplements, we get total com- 
pensation per hour at work, shown in the third column of Table 1 and 
in Chart 1. Total compensation increased from 15.1 cents per hour in 
1889 to $2.24 in 1957, almost 15 times the initial level. For the first nine 
years of the period the trend was slightly downward; the low point was 
13.8 cents in 1898. Thereafter the course of total compensation can be 
broken into five main segments: (1) a gradual rise from 1898 to 1915; 
(2) a very rapid rise from 1915 to 1920 in which money wages more than 
doubled; (3) a drop in 1921 followed by a slow rise to 1930 in which the 
1920 level was not regained; (4) a short, rather sharp drop from 1930 to 
1933; and (5) an almost unbroken rise since 1933, most rapid during 
World War II. Total compensation per hour in 1957 was four times the 
1933 level. 

It seems clear that the pattern of compensation just described is strongly 
influenced by changes in consumer prices, and can be better analyzed if 
we translate it into dollars of constant purchasing power. 



Consumer prices 

The fourth column of Table 1 is an index of consumers prices. For the 
years 1914-1957, it is the familiar Consumer Price Index of the Bureau of 
Labor Statistics, converted to the base 1957=100. To this have been 
linked two similar indexes for earlier periods: my index for the years 
1890-1914, and one year from the index by Professor Clarence D. Long 
which covers the decade of the 1880's. As measured by these indexes, 
the level of consumer prices has tripled since 1889. The cautious wording 
of the last sentence is necessary because the indexes are subject to error. 

There are many reasons why it is hard to measure price changes ac- 
curately over long periods. For example, the indexes do not follow the 
prices of the same bundle of commodities. Many items included in the 
early indexes have disappeared from the present index — such items as coal 
cooking stoves, celluloid collars, or corset covers. And of course, many 

14 



table 1. Manufacturing Wages, Consumer Prices, and 
Productivity, 1889-1957 





CO 


(2) 


(0 


CO 


(5) 

ioiai 

Real 

Compen- 


(6) 

Output 
per 


(7) 
Output per 

Unit of 
Labor and 


Year 


Average 


Wage 


Total 


Consumer 


sation 


Man-hour, 


Capital 




Money 


Supple- 


Compen- 


Price 


(1957 


Manufac- 


Private 




Earningi 


ments 


sation 


Index 


dollars 
per hour 


turing 


Economy 




{Dollars per hr. at 


work) 


(1957=100) 


at work) 


Indexes: 


1929=100 


1957 


2.08 


.162 


2.24 


luOO 


2.24 


213 


179 


1956 


1.99 


.155 


2.15 


96.7 


2.22 


204 


177 


1955 


1.90 


.148 


2.05 


95.3 


2.15 


200 


177 


1954 


1.83 


.138 


1.97 


95.5 


2.06 


193 


168 


1953 


1.81 


.127 


1.94 


95.2 


2.04 


186 


166 


1952 


1.71 


.121 


1.83 


94.4 


1.94 


180 


162 


1951 


1.61 


.115 


1.73 


92.3 


1.87 


177 


160 


1950 


1.46 


.094 


1.55 


85.5 


1.81 


176 


159 


1949 


1.39 


073 


1.46 


84.7 


1.72 


165 


149 


1948 


1.35 


.061 


1.41 


85.5 


1.65 


157 


146 


1947 


1.24 


.059 


1.30 


79.5 


1.64 


152 


143 


1946 


1.08 


.051 


1.13 


69.4 


1.63 


144 


144 


1945 


1.01 


.052 


1.06 


64.0 


1.66 


153 


153 


1944 


1.00 


.047 


1.05 


62.6 


1.68 


152 


148 


1943 


.934 


.041 


.975 


61.6 


1.58 


151 


137 


1942 


.827 


.037 


.864 


58.0 


1.49 


152 


133 


1941 


.701 


.036 


.737 


52.3 


1.41 


152 


131 


1940 


.634 


.036 


.670 


49.8 


1.35 


149 


122 


1939 


.603 


.035 


.638 


49.4 


1.29 


140 


120 


1938 


.603 


.036 


.639 


50.2 


1.27 


128 


115 


1937 


.606 


.027 


.633 


51.1 


1.24 


127 


114 


1936 


.542 


.011 


.553 


49.3 


1.12 


128 


111 


1935 


.537 


.005 


.542 


48.8 


1.11 


126 


106 


1934 


.523 


.004 


.527 


47.6 


1.11 


119 


101 


1933 


.437 


.004 


.441 


46.0 


.959 


114 


91 


1932 


.441 


.005 


.446 


48.6 


.918 


108 


92 


1931 


.499 


.004 


.503 


54.1 


.930 


112 


96 


1930 


.522 


.004 


.526 


59.4 


.886 


106 


96 


1929 


.522 


.004 


.526 


61.0 


.862 


100 


100 


1928 


.513 


not 


.513 


61.0 


.841 


95 


96 


1927 


.510 


available 


.510 


61.7 


.827 


92 


96 


1926 


.503 




.503 


62.9 


.800 


88 


96 


1925 


.497 




.497 


62.4 


.796 


85 


94 


1924 


.498 




.498 


60.8 


.819 


82 


94 


1923 


479 




.479 


60.6 


.790 


76 


90 


1922 


.435 




.435 


59.6 


.730 


77 


85 


1921 


.464 




.464 


63.6 


.730 


70 


85 


1920 


.537 




.537 


71.3 


.753 


62 


81 


1919 


.477 




.477 


61.6 


.774 


63 


82 


1918 


.417 




.417 


53.5 


.779 


64 


78 


1917 


.316 




.316 


45.6 


.693 


65 


73 


1916 


.262 




.262 


38.8 


.675 


69 


77 


1915 


.226 




.226 


36.1 


.626 


69 


72 


1914 


.220 




.220 


35.7 


.616 


62 


70 


1913 


.221 




.221 


35.3 


.626 


62 


76 


1912 


.207 




.207 


34.7 


.597 


58 


74 


1911 


.202 




.202 


34.0 


.594 


51 


73 


1910 


.198 




.198 


33.8 


.586 


53 


72 


1909 


.186 




.186 


32.6 


.571 


52 


73 


1908 


.184 




.184 


32.7 


.563 


48 


68 



15 



Table 1 — Continued 



Year 


U) (2) (3) 


w 


CD 


(6) 


(7) 


1907 


.191 


.191 


33.5 


.570 


49 


73 


1906 


.184 


.184 


32.2 


.571 


50 


74 


1905 


.172 


.172 


31.6 


.544 


50 


69 


1904 


.169 


.169 


31.7 


.533 


49 


67 


1903 


.170 


.170 


31.4 


.541 


48 


68 


1902 


.165 


.165 


30.8 


.536 


48 


67 


1901 


.158 


.158 


30.5 


.518 


46 


70 


1900 


.151 


.151 


30.1 


.502 


43 


66 


1899 


.146 


.146 


29.5 


.495 


44 


65 


1898 


.138 


.138 


29.5 


.468 


— 


64 


1897 


.141 


.141 


29.7 


.475 


— 


64 


1896 


.145 


.145 


29.8 


.487 


— 


60 


1895 


.139 


.139 


30.0 


.463 


— 


62 


1894 


.141 


.141 


30.7 


.459 


— 


58 


1893 


.153 


.153 


32.2 


.475 


— 


59 


1892 


.147 


.147 


32.5 


.452 


— 


62 


1891 


.146 


.146 


32.6 


.448 


— 


59 


1890 


,146 


.146 


32.6 


.448 


— 


59 


1889 


.151 


.151 


33.0 


.458 


36 


56 



A full description of the sources and methods for the new estimates in this table will 
be published separately at a later date. Estimates for 1953-57 are less accurate than for 
earlier years. 

Column 1: Based primarily on data from the Census of Manufactures and the Annual 
Surveys of Manufactures. Interpolating data for 1932-49 are from the Bureau of Labor 
Statistics; 1920-31, from the National Industrial Conference Board; 1890-1919, from the 
bureaus of statistics of Massachusetts, New Jersey, and Pennsylvania; 1889 from Clarence 
D. Long, Wages and Earnings in the United States, 1860-1890 (in press). 

Column 2: Based on published and unpublished data on supplements to wages and 
salaries in manufacturing from the National Income Division, U. S. Department of Com- 
merce, allocated between production and non-production workers on the basis of data 
from BLS Bulletin 1186. 

Column 3: The sum of columns 1 and 2. 

Column 4: 1914-1957, Bureau of Labor Statistics. 1890-1914 from Albert Rees and 
Donald P. Jacobs, Real Wages in Manufacturing, 1890-1914 (typescript, 1958) 1889-1890 
from Clarence D. Long, op. cit. 

Column 5: Index of production from John W, Kendrick, Productivity Trends in the 
United States (in preparation), divided by index of production-worker man-hours at work. 
Hours per man are from the sources of column 1. Production worker employment is from 
these sources and Solomon Fabricant, Employment in Manufacturing, 1889-1939, (New 
York, 1942). 

Column 6: From Solomon Fabricant, Basic Facts on Productivity Change (New York, 
1959) based on Kendrick, op. cit. Output per weighted unit of labor and capital combined, 
private domestic economy. 

important items in the present index did not exist in 1 890 — television sets, 
radios, and automobiles, for example. Even electricity was too unimportant 
to include before World War I. 

Another difficulty is that as we go back in time, price records become 
less adequate. And today it is still hard to measure the true price for 
constant qualities of things whose quality is improving. For example, if the 
price of tires is unchanged, but the number of miles of use in each tire is 

16 



I 



I 






i — r— i 1 1 




*-«» 




w 










1 1 


o 




J - 




.© 








IS 




a 




1- 




£ 




o 




b. 




u.^ 








- S* 




* i 




O 8S 


«o" I \ 


> O 

fa 




S o — 




J «_ «= 


o I 


r i« • 


I \ 


f<3 


*""* 1 


1 4> 


j© \ 


I a 


3 / 


1 C 


£ \ 


\ .2 




^ v» 


■*" I < 


^ c 


*- 1 


v O 


9 *> I 


>v D. 


o o \ 


— ^ £ 


*-Q ) 


i 2 
1 "o 


j= «o 


\ l- 


*.<>» \ 




o «~ \ 




Q. \ 




*= ( 




o \ 








° / 




vt I 




c V 




8. y 




E 




o 




U 




*B 


\ \ 


"5 






jl-Lo 1 ,., l L '. 



o 

CO 



1 

Z 

s 

z 



o *oo o o o 

TOW (S i— 



ooooo o o 

OOC0KS3 V) «* 



o 

CO 



o «o 

CM r- 






increasing, there is a disguised fall in the "price" of tires that may be over- 
looked in the price index. Thus failure to take the improvement in quality 
fully into account will cause us to overstate the rise in the price level. 
To the extent that this is true, we will understate the rise in real wages. 
The general pattern of consumer prices is much like that of money 
wages, but less steep. Both fall until 1898, rise gradually until 1915, and 
rise sharply in World War I. Both drop after the war and then rise again, 
but consumer prices reach their peak sooner — in 1926 rather than 1931. 
The rise in prices after 1933 is much slower than the rise in wages, and 
more of the increase associated with World War II was registered in the 
price index after the fighting was over. 

Real wages 

By dividing the figures on total compensation by the Consumer Price 
Index we get total compensation per hour at work in constant dollars of 
1957 purchasing power. Since this is too large a mouthful to repeat, I shall 
refer to this series simply as real wages. Real wages are shown in column 5 
of Table 1, and in Chart 1. Chart 1 has a logarithmic vertical scale,, which 
gives it the property of always showing equal percentage changes by the 
same slope. 

Even after deflation for the price rise, we find that wages in 1957 were 
almost five times higher than in 1889. This is an astonishing record of 
growth. Although other countries — perhaps Japan and the Soviet Union — 
may have experienced more rapid rates of growth in total output than the 
United States, it is doubtful whether any other country has achieved a 
comparable rate of growth in the real income of wage earners. 

The series for real wages shows three distinct periods. From 1889 to 
1913, real wages rise at the average compound rate of 1.3 per cent an- 
nually. From 1913 to 1929, they rise at the rate of 2.0 per cent annually 
and from 1929 to 1957 at 3.5 per cent annually. Later in the paper we 
shall explore the possible causes of this accelerating rate of growth. 

The sharpest rises in real wages occur from 1915 to 1919, from 1939 
to 1944, and in two jumps during the 1930's: 1933-34 and 1936-37. The 
rises in real hourly earnings in wartime occur partly because of the shortage 
of labor, accentuated in World War I by the cutting off of immigration. 
In World War II, overtime work at premium rates was an important factor 
causing real wages to rise despite wage controls. However, a considerable 
part of the wartime rise in real wages (and consequently, of the postwar 
fall) is probably spurious — a result of measurement errors. 

Despite adjustment for quality deterioration the Consumer Price Index 
probably fails to account for all of the quality deterioration in consumer 
goods during wartime, or for all of the transactions during World War II 

18 



made at black market prices. Thus the index rises too little during 
the wars and too much in the immediate postwar years. More important, 
the use of the Consumer Price Index to measure real wages in wartime 
assumes that workers could spend their whole incomes at the prices given 
by the index. This, however, is not true. There was abnormally high 
saving during World War II (more than one-fourth of disposable personal 
income in some years). In part, people were helping the war effort by 
buying bonds; in part they were saving to buy homes, cars, or other durable 
goods when these again became available; in part they were using some 
of their abnormally high incomes as a cushion against future hard knocks. 
On a smaller scale, something similar must have happened in World War I. 
Had people attempted to spend all their income currently, prices would 
have risen much faster still, and measured real wages would have risen 
much more slowly. 

The first of the two sharp rises in real wages in the 1930's coincides 
with the adoption of the NRA codes, which generally reduced full-time 
hours of work from 48 to 40 per week with no cut in weekly pay. This 
means a rise in hourly earnings of one-sixth. Our money-wage series rises 
a little more than this from 1933 to 1934, while the real-wage series rises 
a little less. The second sharp rise in real wages of the 1930's coincides 
with the recovery of 1937 and with the first major victories of the new 
industrial unions. 

Apart from the two postwar declines, which as has been noted may be 
spurious, the rise in real wages has been remarkably steady. No other 
decline lasted longer than two years; indeed, none since 1908 lasted longer 
than one year. Real wages generally rise in periods when prices are falling, 
such as 1889-90 or 1926-31, as well as when prices are rising. However, 
before 1929 real wages fell rather consistently during the more important 
contractions in general business activity. Such falls occurred in 1894, 
1904, 1908, 1914, and 1921. It is therefore somewhat surprising that in 
the great depression of the 1930's real wages fell in only a single year, 
1932, and did not fall in any subsequent year except for the doubtful 
decline of 1944-46. 

How can one explain this change in the cyclical behavior of real wages? 
It is possible that the change has been brought about by the spread of 
unions, which protect workers from wage cuts in recessions or even win 
them further advances. However, this leaves unexplained the rise of real 
wages from 1929 to 1931, when the fall in employment was precipitous 
and unions in manufacturing were very weak apart from the printing 
trades and the garment trades. 

A simpler and perhaps better explanation is that the change in the 
cyclical behavior of real wages is itself a result of their sharper upward 
trend. The best way of measuring the losses suffered in a recession is to 

19 



measure them relative to the growth that would normally have occurred. 
On this basis, we can as an example compare the recession of 1953-54 
with that of 1907-08. Judged by the year-to-year percentage fall in manu- 
facturing employment, the recession of 1907-08 was the more severe. Real 
wages fell 0.7 per cent from 1907 to 1908, while according to the trend 
for 1889-1913, they should have risen 1.3 per cent. The total gap is thus 
2.0 per cent. From 1953 to 1954, real wages rose 1.0 per cent, when 
according to the trend for 1929-57, they should have risen 3.5 per cent. 
The gap is thus 2.5 per cent — larger than in 1907-08. 

From 1889 to 1957 the average actual workweek in manufacturing fell 
from about 54 hours to about 38 hours, allowing for vacations and holi- 
days. Thus a large part of the gain in real hourly earnings — about 37 per 
cent — has been used to buy leisure during the working years rather than 
to buy goods and services. Since the drop in working hours has been quite 
steady, there is every reason to expect it to continue, though it may take the 
form of longer vacations rather than shorter workweeks as such. 

The historical record: productivity 
Choosing a productivity index 

It has become a commonplace of economic discussion, both professional 
and popular, to say that gains in real wages depend on gains in productiv- 
ity, and that if money wages outrun productivity, prices will rise. There is 
a sense in which these statements are true. Gains in the real income per 
unit of input of any group furnishing inputs to the production process 
must come either at the expense of another such group or as a result of 
getting more useful output per unit of input — from gains in efficiency. 
Since it is difficult for any group to increase its share of income and since 
productivity has been rising rapidly, the second source of gain is generally 
much the more promising. This does not mean, however, that in all cases 
the redistribution of income from existing output is impossible, or wrong, 
or both. 

Similarly, it is true that if money wages rise faster than productivity and 
if shares are constant, prices must be rising. This statement is somewhat 
like the statement that in a set of double-entry books the debits equal the 
credits; it is true by virtue of the way we define the terms, but it is rather 
empty. If money wages rise faster than productivity, they may be pushing 
up prices, or they may have been pulled up by independently rising prices. 
There is no sure or simple way to tell from wage, price, or productivity 
statistics. 

There is a well-established body of economic theory relating wages to 
productivity, but the productivity concept used takes into account both 

20 



the skill of the kind of labor in question and its scarcity or abundance 
relative to other production inputs. Neither of these crucial factors can be 
controlled directly in the productivity we measure by economic time series. 
These cautions should be kept in mind in considering the last two 
columns of Table 1, which show two different measures of productivity 
with some relevance to real wages in manufacturing. Neither is an ideal 
productivity index for our purposes; each has some strengths and some 
weaknesses. Several other possible productivity indexes might be relevant, 
but perhaps confusing as well. 

Output per man-hour at work 

Column 6 of Table 1 shows manufacturing output divided by the number 
of production-worker man-hours worked in manufacturing. This is the 
familiar index of output per man-hour which most people have in mind 
when they speak of "productivity," except that the man-hours are man- 
hours at work rather than man-hours paid for. 

The earnings series of columns 1, 2, and 3 have implicit in them a man- 
hours series. Column 1 is in effect production-worker payrolls divided by 
production-worker man-hours at work. Column 6 is manufacturing pro- 
duction divided by the identical man-hours series. This correspondence in 
coverage and concept with the wage series is the virtue of this particular 
productivity series. In fact, when we compare output per man-hour with 
earnings per man-hour, the man-hour terms cancel, and we are actually 
comparing the movement of manufacturing output with the movement of 
production-worker payrolls. 

It is now generally well realized that output per man-hour does not 
necessarily reflect the contribution of production workers to changes in 
efficiency. It can rise because production workers work harder or are more 
skilled. However, it can also rise because more capital or more non-pro- 
duction workers are used per production worker. It can rise because of 
the improved quality of purchased materials or because of an increase in 
the ratio of purchased materials to final output. And, most likely of all, 
it can rise because of technological change. 

Depending on its source, a gain in output per man-hour may or may not 
imply that real wages should rise. In general, we expect that in the sectors 
of the economy where output per man-hour rises least rapidly, wages will 
outstrip this measure of productivity, so that they will stay roughly in line 
with wages for workers of equal skill elsewhere in the economy. Where 
output per man-hour rises most rapidly, we expect wages to lag behind it. 
Some of the productivity gain will go into lower relative prices, and thus be 
shared with the consumers of the product. It is for this reason that an 
index of productivity for the whole economy may be more relevant to real 

21 



Real Compensation Per Hour At Work, 
Manufacturing, and Two Measures of Productivity 



240 
220 

1 

160 
150 
140 
130 
120 
110 
100 
90 

80 

70 

60 
50 



40 
220 
200 

118 

170 
160 
150 
140 
130 
120 
110 
100 

90 

80 

70 L 

60 

50 



CENTS 



Total Compensation per hour at work, Manufacturing 

Production Workers, 1957 Dollars 

(Table I— Col. 5) 



INDEX NUMBERS 



40 
35 



Output per unit of Capital and Labor Combined, 

Private Domestic Economy (Kendrick) 

1929=100 

(Col. 7) 




Output per hour at work. Manufacturing Production 

Workers, 1929=100 

(Col. 6) 

i i i » i i i 1 — «- 



1890 1900 1910 1920 1930 1940 1950 1957 

Chart 2 



22 



wages in manufacturing than an index for manufacturing alone — for we 
have computed real wages using a price index broad enough to reflect the 
performance of the whole economy. 

Table 1 and Chart 2 show that from 1889 to 1929, output per man- 
hour in manufacturing rose faster than real wages. Since 1929, real wages 
have risen faster, though the changes are almost parallel after World War 
II. If it causes problems for real wages to outstrip output per man-hour, 
the problem period would seem to be 1929 to 1944. Over the whole 
period 1889-1957, the increase in output per man-hour is slightly greater 
than that in real wages. 

Table 2 summarizes some of these comparisons in the form of annual 
percentage rates of change. We shall return to Table 2 after discussing 
the remaining measure of productivity. 

It may be interesting to note that the cyclical fluctuation of output per 
man-hour is somewhat more pronounced than that of real wages, especially 
in the early period. The drop in output per man-hour in recessions occurs 
in part because employers retain more labor than they currently need in 
order not to lose skilled personnel, or they use labor to do maintenance 
work that is not reflected in current output. Then too, the labor force 
cannot be reduced in proportion to output, for certain minimum crews are 
needed as long as there is any production at all. The fall of productivity 
relative to wages in recessions is an important source of the sharp falls 
in profits. 

The index of output per man-hour behaves in a rather peculiar fashion 
in both World Wars. In World War I it falls after 1915, while in World 
War II it is virtually unchanged from 1942 to 1945. These movements do 
not seem entirely plausible. They may reflect in part the difficulty of 
measuring output when its composition is changing rapidly. 

Output per unit of labor and capital 

The last column of Table 1 shows John W. Kendrick's index of output 
per unit of labor and capital combined for the entire private domestic 
economy. As we have seen, a productivity index for the whole economy 
is relevant to real wages in manufacturing for two reasons. First, pro- 
ductivity gains will often be diffused through price reductions and thus 
reflected in the Consumer Price Index. Second, the labor market will 
maintain a rough balance among money-wage rates for work of equal skill 
in different sectors of the economy. It is also meaningful to include the 
input of capital in a productivity index to be compared with real wages. 
If output per man-hour rises because more capital is used, the added output 
is not available for raising real wages until the costs of using the additional 
capital have been met. Output per unit of capital and labor combined is a 
measure of the gains available for distribution. 

23 



Kendrick's productivity index does not use exactly the same measure of 
man-hours that is used in the real-wage series. It is based to a considerable 
extent on man-hours paid for rather than man-hours worked because data 
on man-hours worked were often not available. This causes it to diverge 
further from the real-wage series in recent years than would otherwise 
be the case. 

From 1889 to 1913, real wages rise at the same rate as output per unit 
of labor and capital combined in the private economy, as can be seen most 
easily in Table 2. From 1913 to 1929, real wages rise slightly faster than 
this productivity index, while after 1929 they rise very much faster. Our 
task in the rest of this paper is to interpret these divergent movements in 
the real wage and productivity series. 

table 2. Annual Compound Rates of Growth of Real Wages 
and of Productivity, 1889-1957* 

1889-1913 1913-1929 1929-1957 1889-1957 

Total real compensation per hour at 
work 1.3 2.0 3.5 2.4 

Output per production worker man- 
hour at work, manufacturing 2.3 3.0 2.7 2.6 

Output per unit of capital, manu- 
facturing -1.1 1.9 1.9 0.7 

Output per unit of capital and labor 
combined, private domestic economy 1.3 1.8 2.1 1.7 

* Computed from Table 1, except output per unit of capital, which is from Kendrick, 
op. cit. The 1913 data in this series are interpolated and the last date is 1953 rather than 
1957. The dates have been chosen so as to separate the periods of differing growth rates in 
real earnings and so that each period ends with a prosperous year. 

It may be mentioned in passing that most of the other possible productiv- 
ity indexes that seem relevant to real wages in manufacturing probably lie 
close to or between the two presented here. At any rate, this is true of 
two more of Kendrick's indexes: (a) output per man-hour, all employees 
and proprietors for manufacturing; and (b) output per man-hour, all 
employees and proprietors for the private domestic economy. 



Unit labor cost 

If the hourly compensation of production workers in current dollars is 
rising faster than output per man-hour, the cost of production-worker labor 
per unit of output must be rising. Omitting the man-hours terms, which 
cancel, we can express unit labor cost simply as aggregate production- 
worker compensation divided by output, where compensation is payrolls 
plus the cost of wage supplements. This measure of unit labor cost for the 

24 



period since 1919 is shown in Table 3 and Chart 3, where it is compared 
with the change in the wholesale prices of finished manufactured goods. 

Unit labor cost fell less rapidly than wholesale prices after 1920, though 
by 1929 they had fallen as much. From 1929 to 1940, the over-all changes 
were again the same. Wholesale prices then rose less than unit labor costs 
during World War II, suggesting that price controls were more effective 
than wage controls, or that only the legal prices got included in the price 
index. After 1948, the movement of the two series was roughly parallel, 
though wholesale prices fluctuated more. For the whole period since 1929, 
wholesale prices of finished goods have risen slightly less than unit labor 
costs, perhaps in part because wholesale prices of crude materials have 
risen still less. 

It is clear from Chart 3 that in general there has been a close corre- 
spondence between movements of unit labor costs and of wholesale prices 
of finished goods. Does this demonstrate that wage increases have been 
responsible for price increases? As I have indicated before, not necessarily. 
It may be that rising wages have forced up unit labor costs, which in turn 
have forced up the price of output. But it may also be that both rising 
wages and rising prices are the result of broader inflationary forces — a 
rapidly rising money supply, for example — which increase the demand for 
goods and for the labor to make them. Such rising demand would affect 
unit labor costs and wholesale prices in about the same way at about the 
same time. 

Perhaps we can learn something from the timing of the two series in 
Table 3. If prices rise before unit labor costs, this suggests demand in- 
flation, while if labor costs rise first, this suggests that wages have a causal 
role. Five cycles can be identified with reasonable clarity in each series, 
giving five pairs of peaks and five pairs of troughs. In two of these pairs, 
the movements coincide: the peaks of 1920 and of 1948. In two early 
cases, the movement of unit labor costs appears to precede that of prices: 
the trough in unit labor costs of 1922 seems to correspond to the price 
trough of 1924, and the peak in unit labor costs of 1923 seems to cor- 
respond to the price peak of 1926. In all other cases, the price series leads 
the series of unit labor costs — at the peaks in the price series of 1937 and 
1951 and at the troughs in the price series of 1932, 1939, 1949, and 1953. 
Prices probably lead at the troughs because productivity rises faster than 
wages in the early stages of recovery from recessions; that is, unit labor 
costs fall. 

The timing evidence thus suggests that the common element in the two 
series of Table 3 arises from the demand side. However, annual data are 
extremely crude tools for such an analysis, and this result must be 
taken with great caution. The preparation and analysis of these series 

26 



on a monthly or quarterly basis is beyond the scope of this paper; though 
such series might be helpful, they would not necessarily provide con- 
clusive evidence. 

table 3. Indexes of Unit Cost of Production-Worker Labor 
and Wholesale Prices of Finished Goods, 
Manufacturing 1919-1957 







(1929 


= 100) 










Wholesale 






Wholesale 


Year 


Unit Labor 


Prices of 


Year 


Unit Labor 


Prices of 




Costi 


Finished Goods 2 




Costi 


Finished Goods 2 


1957 


193 


184 


1937 


94 


92 P 


1956 


193 


178 


1936 


81 


87 


1955 


187 


173 


1935 


81 


87 


1954 


186 T 


173 


1934 


84 


83 


1953 


190 P 


172 T 


1933 


72 T 


75 


1952 


186 


174 


1932 


79 


74 T 


1951 


183 


175 P 


1931 


85 


81 


1950 


167 T 


160 


1930 


94 


93 


1949 


168 


157 T 


1929 


100 


100 


1948 


170 P 


162 P 


1928 


103 


101 


1947 


163 


150 


1927 


106 


100 


1946 


150 


123 


1926 


109 


106 P 


1945 


133 


108 


1925 


111 


106 


1944 


131 


107 


1924 


117 


102 T 


1943 


123 


106 


1923 


120 P 


105 


1942 


108 


104 


1922 


107 T 


102 


1941 


93 


94 


1921 


127 


109 


1940 


86 T 


86 


1920 


159 P 


158 P 


1939 


88 


85 T 


1919 


145 


138 


1938 


94 P 


87 









1 Based on the BLS index of production-worker payrolls, wage supplements from Table 1, 
Column 2, and Kendrick's index of output in manufacturing, extended to 1957 by the 
Federal Reserve Board index. The letter P identifies a peak in the series, and the letter T 
a trough. 

2 BLS index of wholesale prices of finished goods converted to a 1929 base. 



Interpreting the record 

The problem restated 

Two measures of productivity have been put forth here as having some 
relevance to the movement of real wages. We find that before 1913, real 
wages closely followed one of these — output per unit of labor and capital 
combined for the private domestic economy. From 1913 to 1929, the cor- 
respondence was not perfect, but was still very closc^ However, from 1929 

27 



to 1944 real wages rose appreciably faster than either productivity measure. 
Since World War II, real wages have closely followed output per man-hour 
of work of manufacturing production workers, and have risen much faster 
than the broader productivity measure. These facts seem to deny that 
there is any simple correspondence between real wages and the kind of 
productivity we have measured. What forces might explain the complex 
pattern that emerges? 

Errors in concept and measurement 

Some sources of measurement error have already been discussed, espe- 
cially in connection with the Consumer Price Index. We should recognize 
that similar errors occur in the production indexes. When the quality of 
output is improving we may fail to account fully for this improvement and 
thus we may understate the rise in output. If the quality bias in the pro- 
duction index is greater than that in the Consumer Price Index, real wages 
will appear to outrun productivity though in fact they do not. But it 
would be extremely difficult to demonstrate the presence, much less the 
size, of this effect. 

On the conceptual side, one of the main problems is the handling of 
taxes in the real-wage figures. The money-wage data include wages used 
to pay income taxes, either by payroll deduction or otherwise. The Con- 
sumer Price Index includes some excise taxes. Net of taxes, money wages 
and consumer prices would both rise less, but the effect on money wages 
would be much larger. Do we want to argue from this that the rise in real 
wages has been overstated? To the extent that workers' taxes have on 
balance been used to subsidize other groups, such as farmers, the rise in 
real wages has been overstated. But to the extent that workers receive 
government services for their taxes, we don't want to deduct taxes in figur- 
ing real wages. Any allocation of government benefits to particular groups 
of citizens must be very tenuous. It seems clear, however, that most of 
workers' taxes are spent in their own interests broadly conceived, and that 
to a first approximation our treatment of taxes has been correct. 



Capital inputs and the return to capital 

Table 2 shows a productivity index that so far has not been mentioned: 
output per unit of capital input in manufacturing. This is also taken from 
Kendrick's work, adapted to fit our purposes. Before 1913 (and until 
1919) output per unit of capital in manufacturing was falling. Output per 
man-hour increased as rapidly as it did only because large additions to 
capital were made. To call forth this additional capital on a sustained 
basis, the costs of using the capital had to be covered before any of the 

28 



added output was available for raising real wages. The increased use of 
capital per unit of output is taken into account when productivity is 
measured as output per unit of labor and capital combined rather than 
as output per man-hour. This thus helps to explain why before 1913 
real wages tended to follow the first of these productivity measures rather 
than the second. 

After 1919, output per unit of capital in manufacturing rose, though 
more slowly than output per man-hour. Since a smaller portion of output 
was needed to cover the costs of additions to capital, more was available 
for raising real returns per unit of labor. 

The gains shown by the index of output per unit of capital and labor 
combined are available for increasing the real wages of labor and the real 
earnings of capital. But since the earnings of capital are measured as 
interest rates or yields, they are already stated in real terms. The rise in 
the price level is reflected in the smaller quantity of physical capital that 
can be bought with a dollar of money capital. But despite rises in the 
average productivity of labor and capital, separately and in combination, 
there has been no upward trend in the return on capital. Indeed, in recent 
years the yields on high-grade corporate bonds have been consistentiy 
below those prevailing in the period 1913-1929. 2 The suppliers of debt 
capital, at least, have thus not shared the fruits of the rise in produc- 
tivity, and the trend for tangible equity capital may well be similar. 
Does this represent the use of power by labor to increase its share at 
the expense of capital? Not necessarily. For the economy as a whole, 
the rising average productivity of capital as measured by the ratio of 
output to capital inputs has apparently been offset by the increase in the 
stock of tangible capital relative to labor input. The contribution of an 
additional unit of capital to output may be no higher today than it was 
in 1913, or it may even be lower. If this is so, there is nothing in 
economic theory to suggest that the return on capital should have risen. 
The full gain in productivity under these circumstances could be distributed 
as higher wages or lower product prices, and could thus show up in a 
real-wage index. The real-wage index would then rise faster than output 
per unit of labor and capital combined, as it has done since 1913. 

Immigration 

The period 1889-1913, in which real wages rose least rapidly, was a 
period of heavy immigration, especially after 1899. There were several 
years in this period in which more than a million aliens arrived in the 



2 For data on the yields to maturity of high-grade corporate bonds for 1913-29, see 
David Durand, Basic Yields of Corporate Bonds (New York, 1942). For more 
recent data, see Moody's Industrial Manual. 

29 



United States. Moreover, most of the immigrants of this period were from 
Southern and Eastern Europe where levels of skill and education were low, 
not only relative to those of the United States, but to those of Britain, 
Germany, and Scandinavia, which had furnished most of the immigrants 
of earlier periods. 

It is thus not surprising that real wages rose slowly in this period, both 
because immigration kept the average skill-level of the work force from 
rising at its recent rate and simply because labor was plentiful relative to 
other resources. Indeed, until recently it has been thought that real wages 
in manufacturing did not rise at all in the two decades 1894-1914. It is 
only the construction of the new and broader cost-of-living index for this 
period which has been used here that reverses this conclusion and produces 
a modest rise in real wages. 

The flow of immigrants was curtailed first by World War I and then by 
restrictive legislation early in the 1920's. This timing coincides well with 
the movements of the real-wage series. It should be noted that the curtail- 
ment in immigration helps to explain not only the change in the rate of 
growth of real wages, but also the change in the rate of growth of produc- 
tivity. The low average skill of the immigrants in the period just before 
World War I and the problems of absorbing them into the work force 
tended to check the rise in output per man-hour. 

The skill and training of labor 

Apart from the cutting off of immigration, broader forces have changed 
the skill composition of labor. New technology has tended to displace un- 
skilled laborers and to require a larger proportion of skilled wage earners. 
This trend is clearly shown in Table 4. Although the table applies to the 
whole labor force and not to manufacturing alone, there is every reason 
to believe that trends within manufacturing have been similar. 

table 4. Male Workers in Selected Major Occupation Groups 
as a Percent of All Male Workers, 1900-1950 





1900 


Gainful Workers 
1910 1920 1930 


Labor Force 
1940 1950 


Craftsmen, foremen, and kindred 

workers 
Operatives and kindred workers 
Laborers except farm and mine 


12.6 
10.3 
14.7 


14.1 16.0 16.2 
12.5 14.4 15.3 
14.7 14.0 13.7 


15.5 19.0 
18.1 20.6 
12.1 8.8 


Total 


37.6 


41.3 44.4 45.2 


45.7 48.4 



Source: Gertrude Bancroft, The American Labor Force: Its Growth and Changing Compo- 
sition, (New York, 1958). Based on data from the decennial censuses of population. 

30 



An increasing proportion of skilled workers among wage earners will 
tend to raise the average hourly earnings of all wage earners quite apart 
from any improvements in technology. Of course, this would not be true 
of an index of wage rates with constant weights for the various skills. The 
effects of the changes in skill-mix on average hourly earnings are consider- 
ably smaller than they would otherwise be because skill differentials in 
wages have narrowed markedly since the beginning of the century. This 
narrowing of skill differentials may itself be an additional cause of the 
increased use of skilled labor. 

The improving skill-mix does not represent an improvement in the 
efficiency with which inputs are used, but represents instead an improve- 
ment in the quality of these inputs. Nevertheless, it is fully reflected in the 
index of output per man-hour, since man-hours are a simple measure of 
labor time unadjusted for skill. Thus the reflection of the improving skill- 
mix in both real wages and output per man-hour may be another reason 
for these series to move closely together in recent years. On the other 
hand, Kendrick's index of output per unit of labor and capital combined 
will reflect little of the change in wage-earner skill-mix for two reasons. 
First, it includes other inputs — capital and the labor of salaried employees 
and proprietors. Second, the labor input component has been weighted to 
remove some of the effects of changes in skill-mix, so as to get an index 
that more nearly measures improvement in efficiency alone. Thus the skill- 
mix factor helps to explain why this index rises less than real wages after 
1919, and especially after 1940, when the proportion of laborers among 
wage earners fell fastest. 

Apart from the change in skill-mix, there has been within each of the 
broad skill categories an improvement in the training of workers, especially 
in their formal education. This factor will also tend to raise both real 
wages and output per man-hour, and will be diluted in its effect on the 
broader productivity index. 

The relative earnings of production and non-production 
workers 

If production-worker wages in real terms have risen more than output 
since 1929, as our measures indicate, then the returns of some other input 
must have risen less. I have already suggested that this is true of capital. 
It is probably also true of non-production workers. Some statistics from 
the Census of Manufactures will illustrate the point. In 1899 the average 
annual earnings of production workers were 40 per cent of those of non- 
production workers. By 1929, they had risen slightly to 49 per cent. After 
1929, the rise was sharper; by 1956 the average earnings of production 
workers were 64 per cent of those of non-production workers. These fig- 
ures are not strictly comparable because the Census of Manufactures has 

31 



considerably broadened its coverage of non-production workers in recent 
years. But the narrowing of earnings differentials is so pronounced that it 
seems most unlikely that changes in coverage can account for all of it. 

Again we may ask whether the gains of wage earners relative to salaried 
workers are the result of economic forces, or are entirely the result of 
changes in power positions. Again there is much to support the former 
view. The rapid spread of public secondary and higher education in this 
century has vastly increased the supply of people qualified to hold white 
collar jobs, at the same time that the principal sources of unskilled manual 
labor were being cut off. These developments tend to raise the relative 
earnings of blue collar workers quite apart from any action by government 
or unions. A contributing factor, in my view, has been the tendency in 
recent years for popular culture to idealize the white collar worker and to 
abase the blue collar worker, thus making blue collar careers less attractive. 

The wage earner's share of income 

It is often considered an established proposition in economics that 
"labor's share" of income is constant through time. This proposition does 
not rest on any theoretical base — it is simply a generalization made from 
observing statistics. For the particular income stream we have been con- 
sidering (the output of manufacturing industry), and for the class of labor 
we have been considering (wage earners), the proposition is not strictly 
true. The first two lines of Table 2 imply that from 1889 to 1929 the 
wage earner's share of output was falling. From 1929 to 1957 it was rising 
again, though by 1929 it had not quite regained the level of 1889. 

The pattern of the wage earner's share implied by our statistics of real 
wages and productivity is the same as that discovered in a direct investi- 
gation of the wage earner's share in manufacturing. Professor R. M. Solow 
of M.I.T. has examined the wage earner's share of value added in selected 
manufacturing industries since 1899, and finds a fall from 1899 to 1929 
and a rise from 1929 to 1951, with the 1951 share slightiy below that 
of 1899. 8 



Collective bargaining 

I have left until last the force in which there is the most interest, the 
spread of unionism and collective bargaining. In general, it is very difficult 
to tell from the kind of record we have been reviewing just what the effect 
of bargaining has been. To do so would require a more precise estimate 



8 See "The Constancy of Relative Shares," American Economic Review, September, 
1958, p. 627. 

32 



than I can make of the size of the effects of other forces discussed previ- 
ously. One episode, at least, seems clearly to be the result of unions — the 
sharp rise in real wages from 1936 to 1937. It is true that 1937 was a year 
of rapid recovery, but despite this recovery there were still about eight 
million unemployed. It thus seems most unlikely that real wages jumped 
upward because of the tightness of labor markets. The wage rise was, 
of course, not primarily the result of bargaining by established unions. 
Much of it was a reaction to the threat of unionization — an unsuccessful 
attempt to head off the CIO. 

Several statistical studies of the influence of unions on wages suggest 
strongly that unionized workers lost ground during the years 1939-1948.* 
In this period the forces of demand inflation pulled up non-union wages 
faster than unions could raise wages in collective bargaining. Since 1948, 
unions may again have been a factor in raising real wages in manufacturing. 
I know of no thorough study of their effect in this recent period, though 
work now in progress may give us further insights. 

In any event, it is important not to take the results of collective bargain- 
ing simply at face value. Unions generally take credit for the whole of 
bargained wage gains, and management often blames them for the whole 
rise in wage costs. The historical record makes clear that real wages in 
manufacturing were rising long before unions were important. Unions may 
have contributed to the acceleration of the rise, but we cannot be certain. 
Most of the rise, however, must be due to technological progress, to the 
growing supply of capital per worker, and to the rising skill and education 
of the work force. 

There is perhaps one other change in the wage-price pattern to which 
unions have contributed. From 1889 to 1898 and from 1926 to 1931, 
gains in real wages come largely from falling consumer prices rather than 
from rising money wages. There have been no such periods since 1931. 
The main reasons for this have been expansionary monetary and fiscal 
policy and the commitment to full employment. However, collective bar- 
gaining creates a strong bias toward taking real wage gains in the form of 
higher money wages. 

In my opinion, this is not an adverse development. It is highly desirable 
for people to have correct expectations of the future price level. Inflation 
or deflation would cause little difficulty if everyone could anticipate them 
perfectly. But it will never be possible for everyone to anticipate a chang- 
ing price level perfecdy. Some decisions to lend or borrow, to be a land- 



4 See for example, Harold M. Levinson, Unionism, Wage Trends, and Income Dis- 
tribution (Ann Arbor, 1951), especially pp. 66-79; Albert Rees, "Postwar Wage 
Determination in the Basic Steel Industry," American Economic Review, XLI (June, 
1951), 389-404, and Stephen P. Sobotka "Union Influence on Wages: The Construc- 
tion Industry,'* Journal of Political Economy, LXI (April, 1953), 127-143. 

33 



lord or a tenant under a lease, or to enter into other fixed money contracts 
will be made on the expectation (for want of a better one) that the price 
level will be stable. If the price level is in fact stable, fewer such decisions 
will prove to be unwise. 

Many people believe that unions do more than make it necessary for real 
wage increases to come in the form of higher money earnings. They believe 
that unions push up money earnings fast enough to force the price level to 
rise. This would be unfortunate if true, but the evidence of this paper is 
not sufficient to test it. In my own opinion, it is not true in any direct sense. 
Perhaps it contains some truth in the indirect sense that unions help to 
create a political atmosphere conducive to inflation. But these are matters 
that I must leave to other contributors to this volume. 



Conclusions 

We have examined the dramatic increases in real wages and in produc- 
tivity of the past seventy years. It is a record that gives us every reason to 
expect continued large gains in real income in the foreseeable future. At 
the same time we have seen that the interrelations between wages and 
productivity have been more varied and complex than is generally believed. 

The advice is frequently given that wage increases should be based on 
gains in output per man-hour for the economy as a whole. For example, 
the Economic Report of the President of January, 1959 states, "Increases 
in money wages and other compensation not justified by the productivity 
performance of the economy are inevitably inflationary." 

The advice that wage increases should be based on gains in productivity 
for the whole economy, like much sage-sounding advice, is far easier to 
give than to take. The difficulties in applying it to any particular situation 
must be as apparent to those who have tried to follow it in wage deter- 
mination as they are to those who have tried to construct accurate measures 
of productivity and wages. One trouble with the advice is that it concerns 
broad averages, and there may be sound reasons why any particular situ- 
ation should deviate from the average in one direction or another. An 
important class of such reasons is shortages or surpluses of labor in 
particular industries or occupations. 

A second difficulty with the statement that wage increases in excess of 
productivity increases are inflationary is that it is true only if there are 
no shifts favorable to wage earners in the distribution of real income. But, 
as we have seen, there have been important shifts in this direction in the 
past, and there may be other shifts in either direction in the future. Sudden 
and temporary shifts in the distribution of income often represent nothing 
more than the use of economic power by an organized group. Gradual 
and persistent shifts, however, are likely to represent the operation of real 

34 



economic forces. They will probably result in a new income distribution 
that contributes more to economic efficiency than would the continuation 
of the old one under changed circumstances, whatever we may think of 
the two distributions in terms of some non-economic standard of justice. 
It is for this reason that it seems unwise to tie wages or any other set of 
payments for inputs to a productivity index. To do so is to freeze the 
distribution of income among groups in the name of checking inflation. 

We must devise methods of checking inflation that give more room than 
this for the apportionment of rewards to changes in the skill and effort of 
different kinds of workers, to changes in the nature of the labor supply, and 
to changes in our needs for labor services. We must devise means of 
checking the use of excessive economic power by labor or business that 
are both more selective and more effective than advice to unions and 
management in general about the just way to set wages and prices. 

In general, the best way of checking monopoly power in product markets 
is to strengthen the anti-trust laws. The problem of checking excessive 
power of labor unions is an even more complex one. It will take creative 
thinking to reach workable solutions. All of the simple solutions, such as 
"putting labor under the anti-trust laws" seem to me as yet to be vague, 
unfair, or inadequate. 



35 



2 




Productivity, costs and prices: 

Concepts and measures 



John W. Kendrick 



Productivity, wages and other costs, and prices are closely related economic 
variables. If consistently defined and measured, the general price level will 
always rise less than the prices of the factors of production (labor and 
capital) in proportion to the increase in total productivity. These terms 
may at first cause difficulties for the non-economist, but their precise defini- 
tion and meaning will be brought out in subsequent sections of this paper. 
A statement about related changes in the variables does not reveal the 
causal forces at work. For example, [if productivity has advanced, wage 
rates and capital return necessarily rise in relation to the general product- 
price level, since this is the means whereby the market mechanism distrib- 
utes the fruits of productivity gains to workers and investors. \ But this 
statement is neutral with respect to the question as to whether product 
prices are pushed up by forces that are able to raise factor prices faster 
than increases in productivity; or whether factor prices are pulled up by 



John W. Kendrick is Associate Professor of Economics at George Washington 
University. Since 1953 he has been a staff member of the Bureau of Economic 
Research; since 1956, Consultant to the Bureau of the Budget. In addition he has 
served with the National Resources Planning Board and in 1946-53 was Acting Chief, 
National Economics Division of the Department of Commerce. Dr. Kendrick is 
author of Productivity Trends: Capital and Labor and National Productivity and Its 
Long-Term Projection. 

37 



forces that inflate monetary demand faster than the physical volume of 
production can be increased by expansion of factor supplies and produc- 
tivity; or whether both sets of forces interact. 

It is not my assignment to analyze the underlying forces which have 
affected changes in product prices, factor prices, and productivity in recent 
economic history. 1 Rather, it is the more pedestrian task of defining the 
concepts and terms, and explaining briefly the nature of the statistical 
measures available to implement the concepts for purposes of analysis. It 
is important that we be clear about the basic concepts we are using in our 
discussions, and understand the meaning and limitations of the statistical 
measures to which we may refer to support or refute explanations of 
inflating forces. 

Despite general awareness of the importance of changes in productivity, 
there is probably more confusion as to the concept, definition, and measure- 
ment of productivity than is the case with respect to wages and prices. A 
major part of this paper is devoted to these matters, since movements of 
productivity index numbers will differ — depending on concept and sources 
and methods used in preparing the estimates of both output and input. 

Labor compensation, or "wages," is a more straightforward concept, but 
here, too, it is necessary to define just what is included, and to indicate 
the sources of data and their reliability. The same is true of nonlabor costs. 
Price indexes are important because (a) they have impact on purchasing 
power of money incomes, and (b) they are necessary for "deflating" the 
current dollar value of national and industry output in order to measure 
changes in physical production and productivity. 

It is essential for analysis that the estimates of the several variables be 
statistically consistent. This matter will be covered when we describe the 
available measures. 



Productivity concepts 

General definition 

The term "productivity" may be denned as the ratio of output to any or 
all of the inputs employed in production. 2 Output and input are measured 
in physical volume terms, since the purpose of productivity analysis is to 
get at the efficiency with which resources are utilized. The outputs, inputs, 



1 See John W. Kendrick, "Trends in Product Prices, Factor Prices, and Productivity," 
Joint Economic Committee Compendium, The Relationship of Prices to Economic 
Stability and Growth (March 31, 1958), pp. 225-236. See also Table 4 below. 

2 For fuller discussion of concept and meaning, see John W. Kendrick, Productivity 
Trends: Capital and Labor, Occasional Paper 53 (National Bureau of Economic 
Research, 1956) and Solomon Fabricant, Basic Facts on Productivity Change, 
Occasional Paper 63 (N.B.E.R., 1959). 

38 



and productivity ratios are generally expressed as index numbers, since the 
chief interest is in their proportionate movements over time. 

The most commonly used productivity measure is "output per man- 
hour." This is a "partial productivity" measure since it is a ratio of output 
to only one class of input, although labor is the most important input. The 
partial productivity ratios are useful in showing economies achieved over 
time in the use of particular inputs. To measure the net saving of all cost 
elements, or inputs, however, and thus the change in overall productive 
efficiency, output should be related to all associated inputs. For example, 
output per man-hour may rise as a result of using more equipment. Only if 
the increase in capital cost is subtracted from the saving in labor cost can it 
be determined if there was a net reduction of costs, and if so, how much. 

Before looking further at the "partial" and "total" productivity measures, 
we must first define output and the associated inputs at both the economy 
and industry levels. 

Output and associated inputs 

The broadest measure of the physical volume of final output of the 
economy as a whole is the real gross national product. It is "real" in the 
sense that the current dollar magnitudes have been deflated by appropriate 
price indexes in order to eliminate the influence of price changes and reveal 
the changes in physical volume of production of the various types of goods 
and services. Because of serious difficulties in measuring the output of 
government which is neither priced nor reported in physical units, the 
broadest measure appropriate for productivity measurement is the real 
gross private product. The values of "intermediate products" (materials 
and services purchased by one industry from another) are included in the 
value of final product (consumer goods, capital goods, and goods sold to 
government), and are not separately counted. 

The inputs corresponding to the national product are the services of the 
basic factors of production, labor and capital (including natural resources). 
Factor services are measured in terms of hours available for productive 
use: (a) man-hours worked in the case of labor, and (b) the hours in 
which machinery and other types of real capital are available if the capital 
factor is included. When man-hours alone are related to output, they are 
generally not differentiated by type of industry or occupation and multiplied 
(weighted) by the average hourly earnings of a "base" period in each. 
However, they may be, and must be if they are to be combined with 
other inputs, since constant dollars then become the common denominator 
of input as well as of a diverse output. 

At the level of the industry or firm, output may be measured gross, i.e. 
in terms of the physical volume of total output; or it may be measured net 

39 



of intermediate products, i.e. as the constant dollar value of gross output, 
less the constant dollar value of purchased materials, etc. The net approach 
is the one consistent with the national product estimates, although in 
practice net and gross output measures show much the same movement in 
nonfarm industries. The inputs associated with industry net output are the 
services of labor and capital employed in the industry. 

The meaning of the productivity ratios 

Ratios of the physical volume of output to any particular input or class 
of inputs such as labor, capital, or materials are useful in showing changes 
over time in requirements for the particular input. This interpretation is 
more clearly indicated when the productivity ratio is inverted; its downward 
movement implies progressive savings or economies in the use of the input. 
For example, an increase in an index of output per man-hour from 100 to 
200 could be expressed as a decline in man-hours per unit of output from 
100 to 50, implying that "unit labor requirements" have been cut in half. 
The partial productivity ratio does not indicate changes in the efficiency of 
the particular input nor of the productive process generally. 

Economies in the use of an input may result from increasing efficiency of 
production generally, or it may reflect changing proportions of inputs (re- 
sulting from changes in technology or in the relative prices of the inputs). 
In the latter case, the use of one input per unit of output may be reduced 
merely because the use of another input is increased. For example, instal- 
lation of new automated equipment may reduce the need for machine 
tenders and other production workers per unit of output, but the require- 
ments for maintenance workers may go up as does the value of equipment 
used in constant prices. In other words, maintenance workers and capital 
have been substituted for production workers. 

In order to measure changes in productive efficiency generally, output 
must be related to all associated inputs. The total productivity ratio reveals 
the net savings achieved in the use of inputs as a whole, and thus the degree 
of advance in efficiency of the productive process. Measures of this sort, 
including capital as well as labor inputs, have been made primarily by the 
National Bureau of Economic Research. Index numbers published by the 
Bureau of Labor Statistics relate to output per man-hour. 

Increases in total productivity reflect primarily 

(1) technological progress resulting from innovations in plant and 
equipment and in the processes or organization of production 

(2) economies of scale, as overhead- type expenses are reduced per unit 
of output and greater specialization of production is made possible as 
output grows, and 

(3) more efficient rates of utilization of capacity. 

40 



Point (3) is primarily a cyclical factor and if productivity comparisons 
are made between years of high-level economic activity (1948-53, or 1953- 
57, for example), average rates of utilization of capacity probably do not 
differ markedly. As already noted, changes in the partial productivity ratios 
reflect factor substitutions in addition to the three variables listed. 

The main element in productivity advance — technological innovation — 
is in part a function of the quantity and quality of resources devoted to the 
discovery and commercial development of improved ways and means of 
production. Research and development outlays have increased steadily in 
relation to sales for some decades, and so have the numbers of scientists 
and engineers as a proportion of the labor force. Behind the emphasis on 
research and development lie basic social values, such as the desire for 
material progress, willingness of the American people to adapt to tech- 
nological change, economic freedom, and the spur of the profit motive in a 
relatively competitive setting. 

Investment in research and development activity results in an increase 
in the real stock of intangible capital represented in peoples' technical 
knowledge, or know-how. Technological advance has also been accom- 
panied by increasing average education and training of the labor force 
required to operate the more and more complex technology. This also adds 
to the stock of intangible capital. In some sense, rising total productivity 
reflects the cumulative result of investments designed to improve the quality 
of tangible inputs, as illustrated by the greater efficiency of a better edu- 
cated labor force. 

Labor productivity measures 

The most usual measures of productivity relate output to part or all of 
the associated man-hours worked. The man-hours are usually a simple 
total, or sometimes each type is weighted by average hourly earnings in a 
base period. Unfortunately, these measures are popularly called indexes 
of "labor productivity." But despite the connotation, they do not measure 
the efficiency of labor in the sense of the effort or skill of this factor alone. 
In fact, changes in labor efficiency as such have probably had a minor 
influence on changes in productivity. 

These indexes also do not measure changes in productive efficiency 
generally, since they are influenced by substitution of capital for labor and 
by savings in materials as well as by technological change. They do show 
savings in labor input achieved over time, better indicated by inverting the 
ratio to read "unit man-hour requirements." Since labor is the major factor 
input, real product per unit of labor input will not diverge greatly from 
total factor productivity measures, especially if the labor measure is com- 
prehensive and weighted by major types of labor input. But it is not as 

41 



good a substitute for total productivity when related to gross output at the 
industry level, since capital and materials inputs may overshadow the labor 
element altogether — as in the petroleum refining industry, for example. 
Several different versions of "labor productivity" measures should be noted. 

Output per Man-hour — This measure is merely the quotient of an index 
of output and an index of unweighted man-hours. We will note several 
variants, but first the use of unweighted man-hours deserves comment. In- 
sofar as there is a relative shift of man-hours toward industries or occu- 
pations with relatively higher rates of pay in the base-period, this really 
means an increase in the quantity of labor input if it is agreed that higher- 
paid man-hours contribute more than lower-paid man-hours to production. 
Since there has been this type of shift, man-hours have a downward bias 
as a measure of "real labor input," and output per man-hour an upward 
bias. Also, from the viewpoint of collective bargaining, increases in output 
per man-hour do not give a measure of the extent to which wage rates of 
various occupational categories can be increased without inflationary im- 
plications, since some increase in average hourly earnings in the industry 
or economy is realized merely as a result of upgrading. 

Certain measures of output per man-hour, including some of those pub- 
lished by the Bureau of Labor Statistics, relate output to production worker 
man-hours only, rather than all employee man-hours or the still broader 
concept of man-hours of all persons engaged including proprietors and 
unpaid family workers. Since World War II there has been a distinct 
decline in production workers as a proportion of all employees or persons 
engaged in manufacturing (where the distinction is particularly relevant). 
Since output results from the efforts of all persons engaged, output per 
production-worker has an upward bias as an efficiency measure. It can be 
argued that some non-production workers, such as those engaged in 
research and development, are not contributing to current output, and that 
insofar as such workers are an increasing proportion, this imparts a down- 
ward bias to output per man-hour measures that include all man-hours. 
Separate data on employment in investment-type activity are not generally 
available, but the defect of the ratio here would seem to be to indicate 
the desirability of including in output an allowance for the real value of 
the capital created by this activity. 

A second problem in man-hour measurement relates to the choice be- 
tween using "man-hours worked" and "man-hours paid for" data. Since 
the number of paid vacations, holidays, sick leave, etc., per employee has 
gradually increased, particularly since 1940, output per man-hour worked 
tends to rise somewhat faster than output per man-hour paid for. The 
divergence is indicated by the alternative BLS estimates for the private 
economy shown in Table 1 . Industry hours data are usually collected only 
on a paid-for basis, but the Census Bureau's Current Population Survey 

42 



data can be used to estimate hours worked — although a portion of the 
difference between the two series presented by BLS may be due to some 
conceptual differences other than the one noted. The consensus is that 
man-hours worked is the proper statistic for purposes of productivity 
analysis. In this case, however, to be consistent with output per man-hour 
worked, average hourly earnings should represent total compensation (in- 
cluding pay for time not worked) divided by man-hours worked. Thus, 
the increasing proportion of time paid for but not worked shows up as 
an increase in average earnings per man-hour worked over and above 
increases in wage-rates as such. 

table 1. Productivity Ratios, Private Economy 1 
Average Annual Percentage Rates of Change 





1889-1957 


1919-1957 


1948-1957 


National Bureau of Economic Research 








Real product per unit of: 








Total factor input 


1.7 


2.1 


2.1 


Capital input 


1.0 


1.3 


-0.2 


Weighted man-hours 2 


2.0 


2.4 


3.1 


Unweighted man-hours 2 


2.3 


2.6 


3.4 


Bureau of Labor Statistics 








Real product per: 








Man-hours worked (Census) 3 






3.6 


Man-hours paid for (BLS) 3 






3.1 



JThe N.B.E.R. estimates relate to the private domestic economy, but real net income 
from abroad, which is deducted from private product to put it on a "domestic" basis, is a 
relatively minor item. 

2 The man-hours used in these estimates are primarily man-hours worked, but in some 
industries only man-hours paid-for data were available. 

3 There are other differences between the man-hour estimates based on Census as compared 
with BLS data. 

Source: Computed by the compound interest formula from the estimates contained in 
Tables 2 and 3. 

Output per Unit of Labor Input (Weighted Man-hours) — When man- 
hours worked are weighted by base-period average hourly earnings in about 
40 industry groups of the private economy, the weighted aggregate rises 
significantly more than straight man-hours, and real product per unit of 
"labor input" (weighted man-hours) rises correspondingly less (see Table 
1 and Chart 1 ) . The weighted measure is economically more meaningful, 
as noted above, since it does not make sense to count an hour worked 
by a highly-trained engineer the same as an hour worked by an unskilled 
laborer. Note, however, that the measure shown does not reflect intra- 
industry shifts among smaller industry classes and among occupational 
groupings. If the effect of these shifts could be measured, weighted labor 

43 



Indexes of Productivity in the United States, 1889-1957 
Estimates for the Private Domestic Economy 



Percentage 
change 
(ratio 
scale) 




+ 


100 


+ 


80 


+ 


60 


+ 


40 


+ 


20 









-10 




-20 




-30 



1889 1899 1909 1919 1929 1939 1949 1957 

Source: S. Fabricant, Basic Facts on Productivity Change, Occasional 
Paper 63 (National Bureau of Economic Research, 1959). 

CHART 1 



44 



input would probably rise significantly more than the available measure, 
and output per man-hour would be subject to even more of an upward 
bias than is indicated by the table. In other words, part of technological 
advance is caused by, or at least accompanied by, an upgrading of the 
labor force which is not fully measured by existing labor input measures. 
To this extent, productivity advance is less than available measures indi- 
cate. Output per unit of labor input also has an upward bias as an 
efficiency measure due to failure to take account of capital, as noted earlier. 

Output per unit of capital input 

This index, while referred to less than output per man-hour, is of interest 
in its own right as well as a component of the total factor productivity 
ratio. Like the "labor productivity" measures, "capital productivity" does 
not indicate changes in the efficiency of capital as such, nor in productive 
efficiency generally. It does indicate economies achieved over time in the 
use of capital per unit of output. The fact that the output-capital ratio 
has generally risen less than the output-labor ratio (Table 1) means that 
capital per unit of labor input has risen, and that the effect of rising 
overall productive efficiency on the output-capital ratio has been offset 
to some extent by the substitution of capital for labor. Conversely, factor 
substitution explains the fact that output per unit of labor input rises more 
than total productivity. 

It is assumed that the input or "services" of capital are proportional to 
the deflated stock of capital goods — land, plant, equipment and inventories 
(net of depreciation in the case of plant and equipment). No adjustment 
is made for hours used as contrasted with hours available, since private 
capital carries an implicit charge, whether in use or not, as opposed to the 
labor factor, which is a direct private cost only when employed. The 
degree of utilization of capital is one aspect of productive efficiency from 
the private enterprise viewpoint; the degree of employment of the labor 
force is an aspect of social efficiency, but most productivity measures are 
constructed from the private economy standpoint. 



Total factor productivity 

Based on capital as well as labor input, this measure has risen less than 
the "labor productivity" measures since total input rose more than 
weighted man-hours due to the growth of capital per unit of labor input 
(Table 1 and Chart 1 ) . Although the best measure of changing productive 
efficiency, total productivity measures are not yet published by government 
agencies (except for agriculture). This is due partly to theoretical and 
statistical problems connected with the measurement of real capital stocks 

45 



and services. But interesting exploratory work along these lines has been 
done at the National Bureau of Economic Research and elsewhere. The 
total productivity studies reveal that output per man-hour indexes for the 
economy and most industries have an upward bias as measures of changing 
productive efficiency. We have already noted that both the labor and total 
productivity measures would probably show still smaller increases if man- 
hours (and capital) could be weighted in detail, and if the services of 
intangible capital were included in the inputs. 

In the absence of total factor productivity measures, and particularly as 
supplements to these measures, indexes of output per man-hour can be 
useful if the hmitations on their meaning are kept in mind. Even as in- 
dicators of productive efficiency, the bias is generally not very large, since 
labor input is the preponderent component of value added. Industries with 
the largest increases in output per man-hour generally show the largest 
increases in total productivity as well. 

But it is important to observe that the bias of labor productivity indexes 
as measures of changes in productive efficiency differs from one period to 
another. For example, in the postwar period 1948-57, output per man- 
hour rose somewhat more than over the long period 1919-57. But this did 
not mean that the rate of advance in total factor productivity had accel- 
erated. It merely reflected a higher rate of substitution of capital for labor 
due to the relatively high rate of investment in the postwar period. The rate 
of advance in total productivity was exactly the same in the postwar decade 
as it was over the longer period 1919-1957 (Table 1). 

Concepts of factor cost or income 

It is useful in analyzing price changes to be able to break down price per 
unit of output into the various types of costs per unit of output. Total costs 
of each type depend on the physical volume of input of the cost element, 
and its price. Thus, changes in labor cost per unit of output, for example, 
depend on changes in output per man-hour and on changes in average 
hourly earnings. If average hourly earnings rise faster than output per 
man-hour, unit labor cost will rise. 

We are also interested in factor cost in its obverse role as factor income, 
and the factor income per unit of input as the price or wage of the factor. 
Wages in current dollars may be deflated by appropriate general price index 
numbers in order to estimate changes in real wages. 

The national income and product accounts provide a consistent frame- 
work for this type of analysis. Corresponding to the notion of the national 
product is the concept of the national income, or "factor cost" of pro- 
duction. The alternative designations merely mean that the employment of 

46 



factor services represents a cost from the viewpoint of the producer, but 
compensation or income to the owners of the factors. For example, wages 
and salaries are the cost of labor to the employer, but income to the 
employees. 

The national income is sometimes called the net national product at 
factor cost. It is net of indirect business taxes and depreciation. It is 
estimated by summing the compensation of employees, net income of 
proprietors, net rents, royalties, and interest, and corporate profits. Func- 
tionally, it may be broken down into the compensations of the two major 
factor classes, labor and capital (including natural resources). 

Labor compensation 

Estimates of employee compensation should include all types of pay — 
not only wages and salaries as such, but also the "fringe benefits" or 
supplements, such as employer contributions to pension funds. The value 
of services to employees, such as the use of health and recreational facilities 
furnished without charge, could be included by imputation, but is generally 
not. Wages and salaries, plus the usual supplements as estimated in the 
national accounts and by Albert Rees for manufacturing, account for most 
of employee compensation. 

To get an average employee compensation per hour, which is the "price" 
of labor services, total compensation should be divided by hours worked. 
This is consistent with the productivity measures, and makes sense in that 
firms are paying for hours of work. A relative increase in hours of paid 
vacation time, sick leave, holidays and other time paid for but not worked 
really represents an increase of labor cost per hour of productive work. 
Divergence between hours worked and paid for has been important only 
since about 1940. In the economy man-hour estimates, I was not able to 
get estimates of average hours worked for all industries, and my series 
include some hours paid for which impart a slight downward bias to the 
productivity and average hourly earnings estimates (but the two series are 
consistent in this regard). The Bureau of Labor Statistics has tried to 
estimate real product per man-hour on both bases (Table 2). 

In studying changes in the functional distribution of income, imputed 
compensation for the labor of proprietors (and their unpaid family mem- 
bers) should be included with employee compensation. The same amount 
should properly be deducted from the total net income of proprietors in 
order to isolate the return to the proprietor's capital and enterprise. Since 
a conventional basis of imputation for the labor compensation of pro- 
prietors is the average earnings of employees in the same industry, the 
average earnings estimates based on employee compensation alone must 
be used in studying movements of the "price" of labor. 

47 



table 2. Indexes of Real Product per Man-Hour for the 
Private Economy, 1947-58 

{1947-49 = 100) 
Man-hour estimates based primarily on data from: 







Bureau c 


f Labor Statistics 




Bureau of Census 








Nonagricultural industries 
























Nonagri- 














Agri- 




Manu- 


Nonmanu- 




Agri- 


cultural 


Year 


Total 


culture 


Total 


facturing facturing 


Total 


culture 


industries 


1947 


96.7 


90.5 


97.5 


97.6 


97.3 


97.4 


90.6 


98.4 


1948 


100.2 


107.1 


99.4 


100.1 


98.9 


100.3 


107.5 


99.4 


1949 


103.1 


102.2 


103.3 


102.6 


103.9 


102.2 


101.6 


102.4 


1950 


110.4 


116.2 


108.8 


109.5 


108.4 


110.3 


116.1 


108.5 


1951 


113.2 


114.6 


110.6 


111.2 


110.0 


115.2 


114.1 


112.8 


1952 


115.7 


124.5 


112.0 


113.0 


111.3 


118.9 


124.0 


115.5 


1953 


120.4 


138.6 


115.1 


118.3 


112.8 


123.9 


138.0 


119.0 


1954 


122.6 


148.3 


116.9 


117.4 


116.7 


127.0 


147.9 


121.8 


1955 


128.0 


153.3 


121.9 


125.6 


120.0 


133.1 


152.9 


127.5 


1956 


128.8 


160.7 


121.8 


127.1 


119.1 


134.2 


160.2 


127.7 


1957 


132.3 


168.6 


124.4 


127.7 


122.9 


137.8 


168.6 


130.0 


19581 


133.4 


190.1 


124.3 


2 


2 


137.6 


190.1 


128.6 



Source: "Recent Changes in Output per Man-hour, Total Private Economy and Major 
Sectors." 

Material submitted by Ewan Clague, Commissioner of Labor Statistics, to Joint 
Economic Committee of Congress. (Bureau of Labor Statistics, processed release, 
1959) 

i Preliminary, subject to revision. 

2 Not available. 

Note: The indexes in this table were computed by the Department of Labor, Bureau of 

Labor Statistics, from estimates of real product and man-hours. The real product estimates, 

referring to 1954 prices, are based primarily on national product statistics of the Department 

of Commerce, Office of Business Economics, except for the manufacturing real product 

estimates which were developed by the Bureau of Labor Statistics. 

Output per man-hour estimates based primarily on Bureau of Labor Statistics man-hour 
data relate, in concept, to man-hours paid whereas estimates based primarily on Bureau 
of the Census labor force data relate, in concept, to hours worked. The difference between 
the two measures may, however, be due in part to statistical as well as conceptual differ- 
ences. Both sets of man-hour estimates cover the man-hours of wage and salary workers, 
self-employed and unpaid family workers. 

Concepts, methods, and sources are described in forthcoming BLS report, "Postwar 
Trends in Output Per Man-Hour, Total Private Economy and Major Sectors." 



Capital compensation 

When labor compensation is deducted from total national income (factor 
cost) for the economy or component industries, the remainder may be 
called the compensation of capital. It consists of net interest, rents and 
royalties, corporate profits, and the nonlabor portion of proprietors' net 
income. This total is useful (a) in surveying the changing distribution of 
national income between the two broad factor classes, (b) for computing 
capital charges (including profit) per unit of output as a companion series 
to unit labor cost, and (c) in estimating "price" of capital which is 
analogous to average hourly earnings. 

48 



Although estimates of real capital stocks and input are in little more 
than an exploratory stage, when capital compensation is divided by an 
index of real capital the quotient may be called an index of the "price of 
capital," that is, compensation per unit of real capital input. During years 
of relatively high-level production, it may be assumed that the real capital 
stock is utilized at comparable rates, so the index reflects compensation per 
real capital hour employed. But in other years, the index would tend to 
reflect cyclical variations in rates of utilization. 

The compensation per unit of real capital input so computed reflects 
changes in two elements: (1) the average price of capital goods, and (2) 
the rate of return on net capital assets. Each of these components, which 
are important for inflation analysis along with average hourly labor earnings 
statistics, can be estimated separately. The average rate of return can be 
calculated as the percentage that capital compensation is of the value of the 
associated net capital assets in current prices. Price indexes are available 
for various major types of capital goods. If capital goods prices rise, the 
absolute return to capital must rise proportionately just in order to preserve 
a constant rate of return on capital values at replacement cost. But if the 
rate of return on capital rises, there is ground for suspicion that price 
inflation may be the result of demand inflation. 

Concept of the price level 

The most comprehensive available index of final prices is the index used to 
deflate the gross national product. We shall call this the "national price 
index." It represents a weighted average of price indexes for all the final 
goods and services purchased in the nation's economy from one period to 
another— consumption goods and services, capital goods, and government 
purchases. Actually, the individual price indexes used for deflating the 
national product are drawn largely from the consumers price indexes (for 
personal consumption expenditures) and the wholesale price indexes (for 
capital goods, including inventory change, and government purchases), 
supplemented by indexes from several other sources, such as the composite 
construction cost index compiled by the Commerce Department and the 
Department of Agriculture's indexes of prices paid by farmers for family 
living. Coverage is comprehensive, but not complete. 

The meaning of the national price index may be expressed by saying that 
it measures changes in the average prices of the changing collection of 
goods and services that make up the national product. This concept may 
be contrasted with that of the Consumer Price Index (CPI), which reflects 
changes in the average prices of the contents of a market basket of goods 
which is held constant over intermediate time periods reflecting the buying 
patterns of urban wage earners in the base period. That is, the national 

49 



price index has quantity weights that change in each time period as ex- 
penditure patterns change; the weights for CPI are changed only occasion- 
ally. The different weighting procedures, as well as the differences in 
coverage, would tend to cause the indexes to move somewhat differently. 

Now, what index should be used for deflating factor income in order to 
estimate real wages, or real nonlabor compensation? Actually, we do not 
have information as to the distribution of each type of factor compensation 
among alternative uses, and so we cannot construct an ideal deflator. The 
compromise that I have adopted is to deflate the compensation of each 
factor class by the national price index. This has the advantage that real 
factor compensation will thereby equal real national product, although the 
deflator does not perfectly reflect the patterns of spending or saving out of 
each type of income. Rees has adopted the more conventional compromise 
of deflating wages (plus supplements) by the Consumer Price Index. This 
is also not perfectly satisfactory, since some wage income is subject to the 
personal income tax and some is saved, although the proportion is probably 
minor. Further, while the CPI is designed to measure the prices paid for 
family living of industrial wage-earners in cities above a certain size, some 
wage income is spent in smaller places, or in types of stores not covered. 

As a practical matter, the CPI and the national price index do not differ 
widely in trend. Thus, the movements of real labor income whether ob- 
tained through deflation by the CPI or the national price index show the 
same general trends. Both are, of course, subject to the shortcomings of all 
available price indexes, which will be discussed further in the next section. 

Sources and methods of measurement 

Not only must movements of productivity, cost and price index numbers be 
viewed in the context of the concepts underlying the measures, but we must 
also remember that the movements are influenced by the adequacy of the 
basic data and the methods of preparing the measures. In general, it may 
be said that all the measures are at best approximations because of lack of 
perfect data, and that their movements are to some extent a function of the 
methods used in their preparation. Here, we can do no more than mention 
a few of the major methodological problems, and give a general appraisal 
of the reliability of the estimates in the light of the nature of the basic 
data sources. 

Prices 

It is convenient to start with the price indexes, since these are not only 
our "dependent variable," but they are used to deflate value of output to 
get real product used in the productivity estimates, and to convert current 
dollar factor compensation to dollars of constant purchasing power. 

50 



All price indexes are based on sample data. It would obviously be im- 
practical and far too costly to collect data on the prices established in all 
transactions in the economy, or even certain major classes of transactions 
such as retail or wholesale sales. The Labor Department collects data for 
certain dates in each month or quarter, from selected distributive outlets in 
selected cities or regions for specified types of goods or services whose 
price movements are believed to be representative of price movements for 
other qualities or price lines of the product type and other types in the 
general product family. In recent years the reasonableness of the "impu- 
tations" involved has been periodically tested, and pricing procedures 
adjusted as necessary. The large number of items priced for both the 
consumer and wholesale price indexes suggests that the indexes reflect quite 
well the broad movement of the general price level that is covered, but this 
cannot be proved since there are not even occasional comprehensive 
estimates. 

The degree of coverage of the broad price indexes has gradually in- 
creased over the years, and the methods employed have improved. 
Although the indexes for earlier years are of somewhat poorer quality than 
those for more recent years, Albert Rees has improved on earlier con- 
sumer price indexes for the years from 1890 to 1914 (when the Labor 
Department index begins) by significantly expanding the coverage of 
relevant items. 

In the overall national product price deflator, there are more imputations 
of price movement of covered to uncovered items than is true of the CPI. 
More importantly, some of the "price" indexes used for deflating certain 
components of the gross private product, and all of the government prod- 
uct, are little more than input price indexes — wage rates, sometimes com- 
bined with materials prices. This type of pseudo-price index is used for 
products that are custom built and therefore cannot be consistently priced 
over time, such as ships, aircraft, and new construction, and also for 
services of households and nonprofit institutions which are obviously not 
priced. Since these input price indexes contain no allowance for produc- 
tivity advance, they are subject to an upward bias to the degree that pro- 
ductivity has advanced in the sectors involved. 

It is well known that price indexes are biased upward to the degree that 
the quality of the same products has improved over time, on net balance. 
Also, since new products are not introduced into price indexes until their 
production is commercially significant, this introduces a slight further 
upward bias into price indexes since usually the relative price of a new 
good declines significantly during the phase of rapid output expansion. 

Finally, there is the weighting problem posed by the fact that the com- 
position of market baskets changes over time, whether it is the market 
basket of the nation or of the families of urban wage-earners. Since there 

51 



is a tendency for purchasers to shift to items which are falling relatively 
in price, recent quantity weights generally result in a slower rise in a 
composite price index than is the case when weights from an earlier period 
are used. The weights of the consumer and wholesale price indexes are 
changed only occasionally, and then the new weights are not used to re- 
weight earlier years but only the time segment from the year of the shift 
onward. The national price index involves annually changing weights in 
respect to the several hundred expenditure categories of the national 
product that are separately deflated, so that it tends to rise less than other 
composite indexes. 

A system of variable weights is required for a price deflator designed to 
convert a value series to constant prices. But even in a consumer price 
index, there are good arguments in favor of frequently changing weights in 
order to reflect the changing patterns of consumption. My own opinion is 
that the CPI contains some upward bias because of the infrequent changes 
in weights. Despite the several sources of upward bias in the available 
price indexes, however, I do not think they are great enough to call into 
serious question the basically inflationary tendencies of our times. 

Productivity 

Evaluation of the sources and methods used to calculate productivity 
indexes must be made in terms of output and input separately. It should 
not be thought, however, that the margins of error in the numerator and 
denominator are compounded in the productivity ratio. Since the data for 
both output and input tend to be drawn from the same or consistent 
sources, and consistent methods are used in estimating both series, it is 
reasonable to suppose that the margins of error attaching to productivity 
changes are less than those attaching to changes in the individual 
components. 

Output — Estimates of the current dollar value of output in the economy 
or its industrial segments, and physical volume estimates that can be made 
for some industries, are tied into data obtained from periodic compre- 
hensive economic censuses (now a quinquennial basis). Estimates for 
intervening years, and for years since the most recently available census, 
are based on sample or partial data. This makes it clear that estimates are 
more reliable in indicating trends over a number of years than annual 
changes; and they are less reliable for years since the most recent bench- 
mark. At present, estimates are being tied into the 1954 censuses; results 
of the 1958 censuses will not be available until 1960. Monthly or quarterly 
estimates are based on still more partial data than the annual estimates. 
Certainly, significance should not be attached to small changes or differ- 
ences in the order of tenths of percentage points. 

52 



Even the trend of the national product estimates prior to 1929 is not as 
accurate as that since 1929, since a number of important economic 
censuses were first taken in that year. Censuses for several industries, in- 
cluding manufacturing, go back to the 19th century, so early output 
estimates for these industries do have reliable benchmarks although the 
degree of product detail is often less in earlier censuses than in those for 
more recent years. 

When physical volume estimates are obtained by deflating the value of 
product, the real output estimates are subject to the same shortcomings as 
are the price indexes. That is, since the price indexes are based on samples, 
their margins of error carry over into the output figures. The use of input 
prices as deflators for some segments of national product creates a down- 
ward bias in real product that I have estimated to be around 10 per cent 
(or 0.2 percentage point) of the average annual secular growth rate. In- 
ability to measure net quality improvements also creates a downward bias 
in real output and productivity corresponding to the upward bias of the 
price indexes. Finally, failure of price data to catch special discounts tends 
to dampen their cyclical amplitude which accentuates the fluctuations of 
real product and productivity estimates. To illustrate, equipment instal- 
lation costs may be absorbed by the seller in recessions but added as an 
extra charge in boom times. Our price indexes do not catch such changes 
in terms of sale. 

Two qualifications with respect to quality bias should be noted. Insofar 
as quality improvement is associated with a larger volume of factor and 
material inputs required by a new model of a product, the associated 
increase in unit cost is deducted from the price change by the Labor De- 
partment, so real product increases to the extent that quality changes are 
accompanied by increased real cost of a commodity. Second, real product 
increases as a result of shifts of expenditures from lower to higher price- 
lines (qualities) of the same product. 

Output is sometimes estimated as a weighted aggregate of physical units 
of various types of goods, such as tons of steel or numbers of tractors of 
various types, as is possible for manufacturing. Theoretically, this would 
give the same result as deflating output values if data were perfect. Since 
data are not perfect, a few possible shortcomings of physical production 
series should be noted. 

Physical unit data are not available for all goods produced. In manu- 
facturing, the proportion of value added covered by items for which units 
were reported increased from about one-half to more than two-thirds 
between 1899 and 1954. The physical volume indexes are usually adjusted 
for changing coverage by a procedure that involves the assumption that 
the average price (or unit value added) of the uncovered items moves with 
that of the covered items. So price imputations, with attendant possible 

53 



margins of error, are involved in "physical" production as well as in de- 
flated value estimates. 

Physical units are often reported for more or less gross categories, and 
not for the various qualities of a product separately. For example, numbers 
of pairs of shoes produced are reported separately for the categories 
"children's," "ladies'," and "men's," but not for the many price lines of 
each. To the extent that there has been a secular shift toward higher 
quality types of a product group, the physical production indexes tend to 
have a downward bias which is not present in deflated value series. But 
since the amount and degree of detail of physical unit data have increased 
in recent censuses, the possible biases have grown less. 

A final point involves the relationship of gross and net output. Most 
industry output measures are available on a gross basis only. In farming, 
for which both measures are available, net output has increased significant- 
ly less than gross because of the increasing relative purchases by farmers of 
intermediate products from other industries. In the case of manufacturing, 
for which estimates of net output have been prepared by the Labor Depart- 
ment commencing in 1947, there has been little difference between move- 
ments of the net and gross output measures. Other evidence suggests that 
this may have been typical of most non-extractive industries. 

Weighting is a technical matter, but it should be noted that recent price 
weights produce somewhat smaller increases in physical output series than 
do earlier weights. The important thing in productivity analysis is that a 
consistent weighting system should be applied to outputs, inputs, and 
related series. 

Inputs — Man-hours worked are usually obtained as the product of 
employment and average-hours-worked estimates. Employment estimates 
have been quite good even on an annual basis since 1939, when Social 
Security data became available. In earlier periods, employment estimates 
are tied into the periodic industry censuses or reports of regulatory agen- 
cies, and in the trade, finance and service areas are tied into industry 
groupings of the occupational data gathered in the decennial Census of 
Population. The estimates give a good picture of trends, but annual esti- 
mates up to 1939 are less accurate since they are based on partial data, 
as are current monthly estimates. It is important that the economy employ- 
ment estimates since 1929, underlying the productivity estimates in Table 
1, are estimated by the Commerce Department on a consistent basis with 
the national product (and employee compensation) estimates. Likewise, 
Albert Rees' estimates for manufacturing are consistent with his output and 
wage estimates, since all are based on the Census of Manufactures. 

The estimates of average hours worked are not as high in quality as the 
employment estimates. Until the postwar period, man-hour data were not 
included in the censuses. For earlier years, and for interpolation between 

54 



censuses, estimates are based on sample or partial data, or adjustments of 
standard workweek data to an actual average weekly hours basis. Much of 
the average-hours data relate only to production workers, in which case it 
is usually assumed that the trend of average hours worked by non-produc- 
tion workers applies to all employees, and to proprietors if they are 
included. Once again, it is apparent that average-hours and man-hours 
estimates are better indicators of trend than of short-period changes. 

Even for trend indications, the tendency toward an increasing proportion 
of time paid for but not worked since 1939 or so has created problems. 
Manufacturing Census and Annual Survey data are on an hours-worked 
basis since 1947, but Rees had to rely on fragmentary data to adjust the 
BLS estimates of average hours paid for to an hours worked basis back 
to the 1930's. The economy man-hour estimates are primarily hours 
worked, but unavailability of necessary information necessitated use of 
hours-paid-for data in some instances, which imparts a slight upward bias 
to the series over the last two decades. 

Estimates of real capital stocks are generally obtained by deflating thr 
book value of assets, or depreciating real purchases of new fixed assets 
over their lifetimes and cumulating the net additions. The latter "perpetual 
inventory" method was used by Raymond Goldsmith in his Study of Sav- 
ings in the United States. Goldsmith's estimates largely underlie the real 
capital input series implied in Table 3. These estimates are not as reliable 

table 3. Productivity Ratios, Private Domestic Economy 

National Bureau of Economic Research Estimates 

(Index Numbers, 1929 = 100) 

Real product per unit of 





Total 








Year 


Factor 


Capital 


Weighted 


Unweighted 




Input 


Input 


Man-hours 


Man-hours 


1889 


56.0 


74.8 


50.0 


43.6 


1919 


82.1 


86.8 


80.4 


79.0 


1929 


100.0 


100.0 


100.0 


100.0 


1937 


113.6 


107.7 


115.6 


114.0 


1948 


145.9 


144.6 


146.4 


156.7 


1953 


166.4 


145.3 


173.1 


190.9 


1957 


179.4 


142 A 


192.6 


211.7 



Note: Annual estimates, plotted in the chart, may be found in Solomon Fabricant, Basic 
Facts on Productivity Change, Occasional Paper 63, National Bureau of Economic 
Research, 1959. 

as the labor input estimates due in part to inadequacies in the capital- 
goods price deflators, and possibly unrealistic assumptions as to economic 
lengths of life of capital goods, not to mention margins of error attaching 
to the current dollar capital outlay or asset estimates. The trends of real 

55 



capital stocks gotten by alternative methods are relatively parallel over 
long periods, but there may be common errors in both methods. 

The same system of occasionally changing weights was used to combine 
individual labor and capital input estimates by industry, and to aggregate 
the two types of input, as was used for the output estimates. As is true 
of consumers, producers tend to shift to inputs that are becoming relatively 
cheaper, which helps account for the increase in real capital relative to 
labor input. Since differences in weight-base make a similar difference in 
the movements of output and of input indexes, the differences between total 
productivity indexes on alternative bases are minimized. 

Factor compensation 

Employee compensation estimates generally come from the same 
sources as the employment estimates, and are of comparable quality. It is 
important that the average earnings and productivity estimates are consist- 
ent as in the economy estimates and Rees' estimates for manufacturing. 

The economy capital compensation estimates are subject to a somewhat 
larger margin of error. Interest and net rent are estimated by a somewhat 
roundabout procedure. But it is the corporate profit estimates that are 
subject to the greatest margins of error as far as the national accounts are 
concerned. Profits are calculated after deduction of depreciation allow- 
ances at original costs, in line with corporate accounting procedure. When 
prices are rising, depreciation reserves are inadequate to cover replace- 
ments at current prices, the level of profits is overstated, and the move- 
ments may be distorted. The Commerce Department is experimenting 
with the estimation of depreciation in constant dollars and current replace- 
ment cost as well as in original cost, which will make possible adjustment 
of the profit estimates, but these are not yet published. 

When compensation is divided by the same output and input indexes 
used in the productivity ratios, the resulting unit-factor-cost and factor- 
price measures are, of course, consistent with the productivity estimates 
and may be used together in price analysis. An example of this use is 
contained in Table 4, which illustrates the basic relationship set forth at 
the beginning of the paper. 



Need for statistical improvements 

Although annual estimates of productivity, prices and wages are available 
going back to the late nineteenth century, this review has indicated that 
they are far from perfect. The estimates probably give a fairly good notion 
of the general order of magnitude of trend rates of change, but even these 
should be interpreted in terms of the known biases in the series — par- 

56 



table 4. Productivity, Product Prices and Factor Prices 
Private- Domestic Economy 

(Average annual percentage rates of change for selected periods) 



Period 


CO 

Product 
Prices 1 


(2) 

Total 

Factor 

Productivity 


(3) 


(4) 

Factor Prices 


(5) 


Totals 


Labor 


Capital 


1919-1953 


1.2 


2.1 


3.3 


3.8 


1.9 


1948-19573 


1.6 


2.1 


3.7 


5.2 


-1.0 



Source: As published in J. W. Kendrick, "Productivity: Contributions of Capital and 
Labor," The Conference Board Business Record, June, 1958. Based on estimates 
by Kendrick for the National Bureau of Economic Research. See also the Joint 
Economic Committee, Compendium, "The Relationship of Prices to Economic 
Stability and Growth" (G.P.O., March 31, 1958), pp. 225-236. 

1 At factor cost. If 100% is added to the average annual percentage rates of change, 
column (1) = column (3) -f- column (2). 

2 Column (3) is the weighted average of columns (4) and (5). Column (4) is based on 
estimates of average hourly labor compensation, and column (5) on the quotient of current 
value capital compensation and the constant dollar stock of real capital goods. In effect, 
column (5) is based on the product of a composite capital goods price index and the rate 
of return on capital. 

3 Preliminary. 

ticularly the downward bias in the real product and productivity estimates 
and the upward bias in the price index numbers. Significance should defi- 
nitely not be attached to small annual changes in the variables. Monthly 
or quarterly productivity estimates, needed for detailed analysis of price- 
cost relationships particularly in current periods, are not available at all 
on a regular basis. 

There can be little question but what further expansion and improve- 
ment in the relevant statistics would improve the quality of our analysis 
and understanding of the inflationary process. A program to improve 
federal statistics was outlined in the 1958 Economic Report of the Presi- 
dent (Appendix C). We shall mention here only a few of the major 
improvements needed in the estimates of the variables with which we have 
been concerned. 

In the first place, the coverage of the price indexes needs to be expanded. 
There are major gaps in the areas of consumer and business services, 
producers goods, new construction, and government procurement items, 
particularly munitions. Further attempts to collect realistic net realized 
prices rather than list prices should be made. Improvement in price series 
not only increases our knowledge of price change, but also of changes in 
production when deflated values are resorted to as in the case of the 
real national product. 

Relatively good estimates of payrolls, employment, average hours and 
average hourly earnings exist for many industries, but even here industry 
coverage needs to be expanded, particularly in the noncommodity-pro- 

57 



ducing sectors. The samples underlying the BLS monthly estimates would 
need to be strengthened in this area. Information about average hours 
worked by nonproduction workers is also needed. Equally important, 
industry data on hours paid for as well as hours worked should be col- 
lected; these would be useful both for productivity estimates and for esti- 
mates of average earnings per hour worked. In addition to direct payroll 
costs, data on supplements to wages and salaries are needed so that total 
labor compensation per hour worked can be calculated. 

With respect to estimates of capital compensation, current quarterly 
estimates of corporate profits and net income of proprietors need to be 
strengthened. Further work in the national accounts is needed to estimate 
depreciation in terms of replacement costs as well as original cost, and to 
adjust the profit estimates accordingly. 

On the output side of the productivity estimates, it will be a major 
contribution when the real private national product can be broken down 
by major segments in addition to the present farm-nonfarm division. Such 
a breakdown within the national accounting framework will permit the 
consistent estimation and analysis of prices, productivity and costs by 
industry. Productivity estimates by smaller industry divisions, such as are 
currently estimated by BLS for several manufacturing and nonmanu- 
facturing industries, should also be expanded. 

Significant improvement in the quality and coverage of output estimates 
depends, of course, on improvements in basic data. The cornerstones of 
our statistical system are the periodic economic censuses, with annual 
sample surveys for intervening years. Yet even the censuses do not cover 
all areas of the economy, and annual surveys are made only for manu- 
facturing and trade. Even within manufacturing, physical volume data 
should be collected in somewhat greater product detail to permit the 
compilation of more accurate output indexes. The coverage of the monthly 
or quarterly Census Facts for Industry reports is far less than that of the 
annual surveys, as indicated by the fact that more than half of the well 
known Federal Reserve Board Index of Manufacturing Production is based 
on adjusted man-hours rather than physical volume series. If we wish to 
know more about production and productivity on a monthly or quarterly 
basis, basic data collection will have to be expanded. There is a limit on 
the extent to which ingenuity can be substituted for data. 

Annual employment data are in good shape, and we have already noted 
the areas in which monthly reporting of employment and average hours 
need to be strengthened. There are, however, no continuing estimates at all 
of the real capital stock of the nation or its industry division. Even if we 
were content to work in terms of output per man-hour rather than estimates 
of total factor productivity, knowledge of changes in real capital per worker 
or per man-hour is important for explaining changes in labor productivity. 

58 



Capital estimates are also necessary to explain changes in capital costs per 
unit of output. It is true that labor represents the major factor cost of 
production, but in analyzing the sequence of price-cost change, labor cost 
must be looked at in conjunction with the cost of capital. 

At best, the analysis of dynamic economic processes is not simple. But 
there can be no doubt that a better analytical job could be done if our 
basic economic statistics were further improved. 



59 



3 



Underlying factors in the 
postwar inflation 




James S. Duesenberry 



Introduction 

Whenever we suffer from any evil, we have a natural tendency to seek a 
devil on which to blame it. The inflation of the last twelve years is no 
exception. Some blame government spending, some blame the trade un- 
ions, some blame the price policies of industry. In fact, the inflation of 
the postwar period is a very complex phenomenon with many causes. The 
rise in the general price level is the resultant of diverse movements of half 
a dozen different groups of prices. Moreover, the forces at work on those 
groups of prices have operated differently in different time periods. 

The complexity of the inflation can be shown by even a very brief 
examination of the movements of different groups of prices. Rents and the 
prices of other services have risen steadily ever since the war. The prices 
of durable manufactures have moved upward in step fashion, periods of 
rapid increase being followed by periods of relative stability. The whole- 
sale prices of non-durable manufactures on the other hand have moved 
erratically. They rose rapidly after World War II, declined moderately 



James S. Duesenberry is Professor of Economics at Harvard University. His books 
include Income Saving Theory of Consumer Behavior and Business Cycles and 
Economic Growth. 



61 



from mid-1948 to the beginning of the Korean War, rose rapidly during 
that war, fell from 1951 to 1956, and rose slowly until the end of 1957. 
Food prices also moved erratically but with even more violent fluctuations. 
Finally retail margins in the durable goods field fell significantly for some 
years after the Korean War. 

The movements of general price indices such as the Consumer Price 
Index (CPI) and the Bureau of Labor Statistics (BLS) wholesale price 
index are weighted averages of the diverse movements of the components. 
The periods of most rapid increase in these general indices are naturally 
those in which all the components were rising. On the other hand, periods 
of stability were those in which increases in some components were offset 
by decreases in others. In particular during the period from mid-1951 to 
mid- 1956, increases in rents, other service prices and wholesale prices of 
durable manufactures were offset by reductions in food prices, prices of 
non-durable manufactures and retail margins on durable goods. The 
general indices conceal as much as they reveal, and we cannot understand 
the nature of the inflation until we understand the causes of the divergent 
movements of the component prices indices underlying the general ones. 

This complex pattern of movements (and I have drastically understated 
the real complexity in the interest of brevity) should make one suspicious 
of any simple general explanation of the postwar movement of the 
price level. 

In this paper I have attempted to sort out the various factors responsible 
for the rise in prices since the end of World War II. To that end I shall 
begin with a brief review of the classical or "demand-pull" explanation of 
inflation and its newer competitor the "cost-push" theory. Before attempt- 
ing to evaluate those explanations, we must examine in some detail the 
movements of different types of prices. Later, the movements of several 
different types of prices since the start of the Korean War are examined 
in some detail. It is argued that the major factor responsible for the general 
rise in prices since 1951 is the high rate of increase in wages generally. 
The effects of rising labor costs were offset for a time by the downward 
adjustment in profit margins, and raw material prices from the high levels 
reached during the Korean War. Rising wage costs were also offset for a 
time by falling retail margins on durable goods. From the beginning of 
1956 the effects of rising wage costs were reinforced by the cyclical upturn 
in profit margins and by rising raw material prices. Throughout the post- 
war period, prices of consumer services, particularly transportation and 
medical services, have been rising more rapidly than other prices. This was 
largely due to their slow rise during the war. 

Some temporary factors like lagged adjustment of service prices have 
reinforced the effects of rising wage costs, and others like retail margins 
for durables have offset them. 

62 



Similarly, profit margins and raw material prices have moved erratically 
— sometimes reinforcing, sometimes offsetting rising wage costs. Those 
factors obscure our understanding of the picture, but it seems safe to say 
that over the whole period since the Korean War their effects on the 
general price level roughly cancel out. We are left with the conclusion 
that rising wages are primarily responsible for the rise in prices since 1951. 

That does not tell us what caused the inflation, it only shows that the 
causes worked through wages. In another section we turn to the causes 
of the wage inflation. It is argued first that neither the "trade union pres- 
sure" explanation nor the "demand-pull" explanation is by itself adequate 
to explain what has happened to wages. It is then argued that demand- 
pull has been a major factor in the rise in wages in the trade and service 
sectors, that trade union pressure is the more important factor in manu- 
facturing, construction utilities and transport. Finally, it is argued that 
trade union pressure and the balance of supply and demand in labor 
markets interact with one another in determining both the extent of wage 
increases in unionized industries and the secondary effects of those in- 
creases on wage movements in other sectors. 

Trade union pressure played a major role in the inflation of the past 
few years, but that pressure would have been much less effective and wages 
of unorganized workers would have risen still less had demand for labor not 
been so strong. We turn later to the explanation of demand and argue 
there that the special buoyancy of postwar demand has been due to 

(1) the very favorable situation in which the economy found itself at 
the end of the war, 

(2) the high marriage and birth rates of the postwar years, and 

(3 ) the rise in government expenditures. 

Many other factors of course influence the movement of income, but those 
are the ones which account for the strength and duration of the post- 
war boom. 

There is, of course, no particular reason for picking out any one of 
those factors as responsible for the demand component in this inflation. 
If any one of them had been removed, we would have had no problem of 
excess demand for labor. 

Finally, it must be noted that the supply of labor (excluding women 
over 45) grew at a very low rate in the postwar years. To some extent our 
problem was not an abnormally high rate of growth of demand but a low 
rate of growth of labor supply. 

I shall conclude with a few remarks on the policy problems posed by 
this analysis of the causes of inflation. 



63 



Theories of inflation 

The classical theory 

Inflation is not a new phenomenon, and economists have long had 
explanations which seemed to give a satisfactory account of the observed 
periods of inflation. Most of those incidents, it should be noted, took 
place during wars or immediately after wars. The classical analysis of 
inflation can be summarized in the following way. 

Suppose that for any reason the aggregate amount of goods and services 
(measured by their value at current prices) which households, business and 
governments wish to buy increases from one year to the next. Suppose 
that the increase exceeds the increase in industrial capacity. But also sup- 
pose that additional labor can be recruited without an increase in wage 
rates. As a result, prices and profit margins will tend to increase because 
additional output can only be obtained at higher cost, through overtime 
work, use of obsolete plants, or badly located plants, etc. That tendency 
may be offset by the increase in labor productivity which takes place with 
the passage of time. In that case the increase in demand has forestalled 
a decline in prices instead of causing an increase. How much of a price 
increase will result will depend on competitive conditions and pricing 
policies of firms and on the speed tvith which the increase in demand 
takes place. 

In industries dominated by a few large firms following long-run price 
policies, the increase may be confined to the elimination of price shading 
of various kinds because those firms expect capacity to catch up with 
demand and do not wish to first increase prices and then reduce them. In 
sectors with larger numbers of firms, prices will have to rise enough to make 
it currently profitable to produce the additional output. 1 In agriculture, 
prices will have to rise enough to bring demand into line with available 
supplies because of the time lag involved in increasing output. The amount 
of price increase will be much influenced by the expectations of speculative 
holders of existing stocks. In any case, it is clear that any increase in de- 
mand in excess of the increase in capacity will tend to produce price 
increases. 

Let us now revise our assumptions and suppose that industrial capacity 
increases as much as the amount of goods demanded at the initial price 
level. Suppose, however, that there was no more than the frictional mini- 
mum amount of unemployment at the start and that the increase in output 



1 In some cases it may be physically impossible to increase output to meet demand. 
In competitive industries prices will then have to rise enough to reduce the amount 
demanded to the amount available. In industries dominated by a few firms there 
may be informal rationing or a build up of unfilled orders. 

64 



potential of the labor supply (resulting from increased labor supply and 
increased productivity) falls short of the increase in demand. In those 
circumstances wages will tend to rise as employers bid against one another 
for labor (if the increase in demand is regarded as permanent). Prices will 
also be increased. In industries with a small number of firms, prices are 
likely to rise by the full amount of the increase in costs (including the 
increase in raw material or component costs due to rising wages in other 
industries). That is so because prices in those industries are limited by 
such factors as fear of attracting entry or fear that customers will decide 
to make rather than buy. 2 If production costs rise for old firms they will 
rise for new ones as well. 

In other industries there may be some squeeze on profit margins, but 
most of the increase in wage costs (plus increases in materials costs) will 
be reflected in prices. 

We have considered separately the cases in which demand increases 
faster than industrial capacity or faster than the output potential of the 
labor supply. Of course, both limits on output may be effective at once. 

Thus far we have considered a one-shot increase in demand. What hap- 
pens next. The mere fact that prices have risen will not stop the inflationary 
process. 3 The increase in prices would reduce the amount of goods and 
services demanded if money incomes did not rise, but it is in the nature of 
the case that money incomes and expected future incomes will rise with 
prices. If wages and prices rise, consumer expenditures will increase. On 
the other hand, when there is a price increase but no wage increase (i.e. 
when demand increases faster than capacity) profits are increased and 
businesses are likely to expand investment expenditures by at least as much 
as the increase in profits. After the first round of price and/or wage in- 
creases, real demand (i.e. expenditures adjusted for price change) is likely 
to be just as high as before the price increase. 4 However, if the initial force 
behind the first increase in demand was a single shot, real demand will not 
increase and, after some further price and/or wage increases, capacity and 
labor supply will eventually catch up and stop the price increase. Accord- 
ing to classical theory, a continuing inflation will take place only if demand 



2 This is, of course, a crude and inadequate statement of the many factors determin- 
ing prices in oligopolistic enterprises. The important point is that firms in these 
industries have not usually charged the prices which would maximize their profits 
in the short run. They are therefore able to raise prices when costs rise. 

3 In the discussion which follows it is assumed that the terms on which credit is 
available are not changed by the rise in prices. Monetary factors are discussed in a 
later section. 

4 For purposes of this discussion, I have assumed that the government's budget for 
purchases of goods and services is fixed in real terms in that money expenditures for 
goods and services will rise with the price level. With fixed legislation some types 
of tax revenue will rise by a greater percentage than the price level. Other types of 
tax revenue and transfer payments will rise less than proportionately. Some rough 
calculations indicate that the two factors will just about cancel each other. 

65 



at constant prices tends to grow faster than industrial capacity or labor 
supply, adjusted for productivity increase. 

The cost-push theory 

The so-called cost-push theory is not strictly speaking an entirely inde- 
pendent theory of the causes of inflation. It is rather a major amendment 
of, or addition to, the excess-demand theory outlined above. No one who 
supports the cbst-push theory denies the essential correctness of the argu- 
ment given above. Those who support the cost-push approach deny the 
applicability of the classical theory to the inflation in the American econ- 
omy in the last few years. Essentially, they argue that trade union pressure 
can cause prices to rise in circumstances in which the classical theory tells 
us they ought not to rise. 

Suppose for example that there is moderate unemployment so that there 
is no shortage of labor at existing wage rates but no great surplus either. 
Similarly suppose that industrial capacity is fairly well utilized but not 
excessively so. In those circumstances employers have no incentive to raise 
wages or prices. If, over a period of time, demand increases at about the 
same rate as labor supply (adjusted for productivity) and industrial ca- 
pacity, prices will be constant or perhaps fall as productivity increases. 

On this peaceful scene the villain now appears — the trade union. Trade 
unions, by threatening to strike, slow down, and raise hell generally, force 
employers to grant wage increases which they would not otherwise give. 
The employers then raise prices by the corresponding amount or propor- 
tion. If we can assume either (a) that trade unions have organized the 
whole labor force or (b) that they exercise an indirect influence on the 
wages of unorganized workers, so that their wages also increase, the general 
rise in wages does not cause unemployment. Everything is essentially as it 
was before, except that prices and wages are higher. 

Assumption (a) is obviously untrue for the American economy and 
assumption (b) is extremely doubtful. Alternatively, suppose that demand 
for the commodities produced by trade unionists is either inelastic (sales 
do not fall much as prices increase) or growing. Then it can be argued, 
though I shall not go into details, that the wage and price increases are 
likely to generate roughly as large a percentage increase in money expendi- 
tures as in the average price level. The process is not per se self-limiting 
in the short run. 

Over long periods there must be some limits to the power of trade 
unions to push up wages relative to the wages in the unorganized sections. 
Presumably, as union non-union differentials increase, the danger that non- 
union firms will grow at the expense of unionized ones will increase and so 
will the danger that new non-union firms will appear. Similarly, the effect 

66 



of competition from substitute products produced by non-union workers 
will become greater. The effectiveness of these factors in limiting wage 
increases will vary with the competitive structure of the industry, ease of 
entry, and the ability of the trade union to restrict the activities of non- 
union workers and employers. 

The effectiveness of those limiting factors may be reduced if increases 
in union wages have a significant influence on non-union wages. However, 
it does seem likely that unions whose membership covers only a fraction 
of the labor force cannot push up wages indefinitely unless there are other 
factors which tend to generate wage increases in other sectors. 

The monetary factor 

An inflationary spiral, whether originally due to excess demand or to 
trade union pressure, may in principle be choked off if the money supply 
does not increase. If prices increase and the volume of real production 
does not decrease, the amount of money balances which businesses and 
households wish to hold will increase. This will cause interest rates to rise 
and may also reduce the availability of credit (i.e. cause lenders to refuse 
to lend to some borrowers of marginal quality who were previously 
accommodated) . 

Prior to 1951 the monetary factor did not impede the inflation, because 
the Federal Reserve was committed to a policy of providing enough bank 
reserves to prevent a long-term rise in interest rates. But from the end of 
1952 to the end of 1957 demand deposits and currency have grown by 
only 4%, while during the same period Gross National Product (in 
dollar terms) has grown by about 40%. Interest rates have accordingly 
risen. However, the extent of the rise required has been limited by the 
willingness of the public to hold money substitutes, i.e. time and savings 
deposits and savings and loan shares and treasury bills. The resulting rise 
in interest rates and the accompanying reduction in availability of funds to 
borrowers of moderate quality (and to those trying to obtain government 
guaranteed mortgages) have undoubtedly restricted demand to some extent 
but not enough to control the inflation. 5 

Price movement since 1 950 

As I have already indicated, the prices of different types of goods have 
followed different patterns in the postwar years. To gain an understanding 
of the movements of the general price level, we have to have some knowl- 



5 It is undoubtedly true, however, that if the Federal Reserve wished to adopt a 
sufficiently stringent policy, it could limit the growth of money expenditures to any 
extent that it wished. 

67 



edge of how its components moved. In this section I shall consider the 
movements of a number of different kinds of prices in the period since 
the beginning of the Korean War in June 1950. The movements of prices 
in the years from 1946 to 1950 were so dominated by the special circum- 
stances connected with decontrol and reconversion that they throw little 
light on the long-term inflation problem. That is also true of price move- 
ments during the first year of the Korean War. But the latter movements 
left some important after-effects we must consider if we are to understand 
later price movements. 

In this section we will first consider the reasons for the rapid rise in the 
prices of services in the years since World War II. There follows a brief 
outline of the movements of raw materials prices since the start of the 
Korean War and a very brief comment on the decline in retail margins 
for consumer durables. Next there are some observations on the changes 
in labor cost in manufacturing and a discussion of manufacturing profit 
margins. Finally, the results of the preceding section are brought together 
in a discussion of their influence on the prices of finished manufactures 
and on the Consumer Price Index. 



Service prices 

One of the outstanding features of the pattern of price movements since 
the end of World War II is the steady upward movement of the prices of 
services — gas and electricity, transportation services, medical services and 
personal and repair services. Other prices have had their ups and downs 
and periods of stability, but service prices have moved upward month 
after month and year after year. Moreover, since 1951 service prices as 
a whole have moved upward more rapidly than the general price index, or 
than any major category of prices. Indeed, since 1951 virtually the whole 
increase in the Consumer Price Index is due to the increase in the prices 
of services and rents. The BLS All Services Index rose from 114in 1951 
to 140 at the end of 1957. The commodities index rose from 110.3 to 
114.7 (see chart: Prices of Commodities and Services). 

Further examination of the price indices for services indicates that be- 
tween 1939 and 1948 the prices of medical services, transportation 
services, and gas and electricity rose much less than the prices of most 
other things. Other services (consisting mainly of laundry, dry cleaning, 
barber and beauty shop services and various kinds of repair services) 
rose in price by about as much as most commodities. From 1939 to date, 
however, the prices of services as a whole have risen only a little less than 
the prices of commodities. (The rise was much smaller for gas and 
electricity, much larger for transportation, and somewhat larger for medical 
services and other services.) 

68 



Finally, wage increases in the service industries were not very different 
from those in manufacturing, though they were slightly smaller. We do not 
have adequate figures on productivity in the service field. However, esti- 
mates of productivity-change for the private non-agricultural sector as a 
whole show a somewhat smaller increase in productivity than the estimate 
for manufacturing. Wage costs for the service industries as a whole must 
therefore have changed in roughly the same way as for manufacturing. 

Over the whole period since 1939, then, service prices have moved 
upward with wage costs. Electricity rates rose much less than other prices 




69 



partly because labor costs are less important than in other industries 6 
and partly because of the great improvement in load factors. Public trans- 
portation prices have risen more than commodity prices because of the 
volume problems of the public transport services. 

The relative rise in service prices since 1951 is due to two factors. 
First, the prices of public transportation, gas and electricity rose less than 
other prices during and immediately after the war and more afterward 
because the regulatory process works slowly. Prices of medical services 
followed the same pattern because of the delaying influence of custom on 



6 This is of course only a temporary factor whose influence will decline as old capital 
is replaced with capital whose price is based on postwar labor costs. 



Consumer Prices on a Prewar Base 

Index: 1939 = 100 



220 




I 
'41 



T 

»43 



-200 



-180 



-160 



-140 



-120 



100 



1939 '41 '43 '45 '47 '49 '51 '53 '55 '57 

Sources: Departments of Commerce and Labor and Committee for Economic Development 

70 



changes in medical charges. Second, commodity prices have been held 
down since 1951 by the decline in agricultural and raw material prices 
which do not influence service prices. 

Agricultural and other raw-material prices since 1951 

In the period since 1951, price-level movements have been influenced to 
an important extent by the movement of the prices of agricultural products 
and other raw materials. 

Between 1951 and 1956, the prices of farm products fell by 23 per cent 
and although there has been some recovery since 1956, the farm price level 



Wholesale Prices of Producer and Consumer Goods 
and Construction Costs 




Index 1947-49 = 100 



—I 

1951 



S 



~T — 
1947 



1955 



1949 1951 1953 

Sources: Departments of Commerce and Labor and Committee for Economic Development 



1957 



71 



at the end of 1957 was still 19 per cent below the level of 1951. The 
general downward movement is due to two major factors: 

(1) The price increase during 1950 resulted from a very rapid increase 
in demand over a period so short that increased prices could not influence 
supplies (with some assist from speculative factors). As soon as farmers 
had time to respond to higher prices (and the rate of increase of demand 
was reduced) supplies increased and prices were forced down. At the 
same time the effect of speculative factors was reversed. It should be noted 
that in the case of meat animals, the effect of a change in prices on supplies 
is spread over several years because of the long period required to produce 
meat animals. 



Wholesale Prices of Durable and Nondurable Manufactures 
and Raw Materials 



150' 



Index 1947-49 




Sources: Departments of Commerce and Labor and 



Committee for Economic Development 

72 



: 



(2) Since the war, agricultural productivity has been growing very 
rapidly, and farm output has generally tended to rise more rapidly than 
demand in spite of a substantial decline in the farm labor force. This 
longer-period force works to prolong the decline in agricultural prices 
initiated by the "kickback" from the Korean War boom. 

The partial recovery of farm prices after early 1956 was due at least in 
part to the depressing effect of low prices on farm output. Because of the 
lag in the response of farm output to price changes, supplies of farm prod- 
ucts began to decline just when consumer demand was rising rapidly. 

The prices of crude non-agricultural materials followed a pattern similar 
to that of agricultural materials up to 1954. The price index for those 
materials rose by more than 30 per cent from 1949 to 1951, then fell 
almost to the pre-Korean level by 1954. Though they rose by 10 per cent 
from 1954 to 1956, they have since fallen slightly. 

The prices of these materials responded to the Korean War in much the 
same way as agricultural materials. But after 1954 they moved differently 
because (a) there is no long-term force tending to depress their prices, 
(b) the fluctuation in durable-goods demand from the beginning of 1955 to 
date has been much more violent than the fluctuation in consumer demand 
for food and other consumer non-durables. 

Retail margins on durable goods 

From early 1951 until near the end of 1955 the wholesale prices of 
consumer durable goods rose by about 5 per cent. Meanwhile, retail prices 
of consumer durables fell by about 10 per cent. From the beginning of 
1956 both price indices rose, but the retail index rose less than the whole- 
sale index. During the whole period, then, the retail margins on durable 
goods declined to a marked extent. This was apparently due to the wide- 
spread development of discount houses selling at low margins in the house- 
hold appliance field and to a general reduction in the margins on sales 
obtained by automobile dealers. 

Wages and labor costs in manufacturing 

The outstanding fact about wages and labor costs per unit of output in 
manufacturing is that they have risen throughout the whole postwar period. 
That is true by almost any measure of labor costs. But the extent of the 
rise depends a good deal on how labor costs are measured. Before any 
figures are cited, a few comments on measurement problems are in order. 

The most commonly cited measure of wages is that for "average hourly 
earnings." Data for average hourly earnings are available by finely classi- 
fied industries monthly. The figure measures pay received for hours 
worked, including premium payments for overtime. To exclude the effects 

73 



of cyclical variations in the amount of overtime, the BLS also publishes a 
series for Straight Time Hourly Earnings, which is obtained by adjusting 
the Average Hourly Earnings Series. Neither of these series measures the 
full costs of labor to the employer. They do not include pay received for 
hours not worked, i.e. mainly paid holidays and vacations. Those costs are 
included in data for "payrolls." Payroll costs per hour worked can be 
computed by dividing the payroll indices by man-hours indices. 

In addition to payments for hours not worked, employer labor costs 
include payments for "fringe benefits" such as pensions, social security 
taxes and supplementary unemployment benefits. 

These latter items are included in the Commerce Department series on 
compensation of employees. Unfortunately, the BLS data on man-hours 
cover only "production or non-supervisory employees." One cannot there- 
fore compute directly a series for total compensation of employees per 
man-hour. 

These considerations are important because payments for paid vacations 
and holidays and fringe benefits have been increasing faster than regular 
wage payments and because "non-production" workers have increased as 
a percentage of total numbers employed in manufacturing. 

The best measure of labor cost seems to be obtained by dividing com- 
pensation of employees in manufacturing by the index of manufacturing 
production. This measure is available only for all manufacturing and on an 
annual basis, and it does not permit us to consider the effects of changes 
in compensation rates and in productivity separately. For those purposes 
we must use other measures. 

In manufacturing as a whole, compensation of employees per unit of 
output increased by 12 per cent from 1951 to 1957. The increase was 
erratic, but compensation per unit of output increased in every year except 
1955. During the same period production worker payrolls per unit of 
output increased only 4 per cent. The difference was due to the relatively 
rapid growth in the number of non-production workers and fringe benefits. 

Profit margins in manufacturing 

In discussing the classical or demand-pull approach to the inflation 
problem we noted that an increased demand relative to industrial capacity 
will tend to raise profit margins and prices. Profit margins will tend to 
rise for two reasons: (1) because prices rise faster than wage costs at a 
given volume of output, and (2) because unit overhead costs (for de- 
preciation, interest, property taxes and overhead labor) fall with rising 
volume when output grows relative to capacity. 

The first cause of increase in profit margins is inflationary. The second 
is not. During the first few months of the Korean War, wholesale prices 

74 



of manufactured products rose faster than production-worker payrolls per 
unit of output, but from June 1950 to March 1951, wholesale prices of 
manufactured goods and production-worker payrolls per unit output had 
both risen by about 12 per cent. Since raw-material prices had risen by 
more than 12 per cent it appears that wholesale prices of manufactured 
goods rose by a smaller percentage than the increase in direct costs of labor 
and materials per unit. Even in the first year of the Korean War, then, 
prices did not rise because of an increase in the margin between prices 
and direct costs. 

However, manufacturing firms did benefit greatly from the improvement 
in volume. Between 1949 and 1951, unit value added in manufacturing 
(which measures the wholesale prices of manufacturing products adjusted 
for changes in the prices of raw-material inputs) rose by about 12 per cent 
while total compensation of employees rose by only 8.5 per cent. Since 
prices rose about as much percentagewise as direct costs, the difference 
represents the effect of spreading overhead over a larger volume. 

Between 1949 and 1951, bef ore-tax profits in durable manufacturing 
increased from 13.4 per cent of sales to 15.9 per cent. In non-durables, 
the increase was from 14.5 per cent to 18.2 per cent. Because of the 
increase in corporate income taxes, after-tax margins actually fell in both 
cases. How much influence the change in corporate income taxes had is 
problematical. 

After 1951, profits of manufacturing corporations as percentages of 
sales began to decline. They reached their low, of course, in 1954 and 
recovered in 1955 and 1956. Even in 1956, however, they had not recov- 
ered to the 1949 level. Allowing for the rise in prices, profits in dollars per 
unit of physical output were just about the same in 1956 as in 1949. Over 
the period as a whole, then, prices were substantially unaffected by changes 
in profits. 7 But changes in profit margins did contribute appreciably to the 
upward movement of prices during 1950 and again in 1955 and 1956. On 
the other hand, declining margins offset other inflationary factors from 
1951 to 1955. 

It seems to me that the movements of profits correspond to those which 
would be expected from the classical model on the assumption that (a) 
over the period as a whole, real demand grew slightly less rapidly than 
industrial capacity and (b) it grew more rapidly during 1950 and again 
during 1955 and 1956, falling behind again in 1957. 



7 Over the period, however, depreciation allowances did rise relative to replacement 
costs. But even after allowance for that consideration, the net effect of profits on 
prices from 1949 onward was negligible. 

75 



Prices of manufactured products 

Wholesale prices of all finished manufactured goods rose rapidly from 
1949 to 1951 when raw-materials prices, profit margins and compensation 
of employees per unit of output were all rising. After 1951, wholesale 
prices of finished manufactured goods actually fell until 1955 by about 
one per cent. During that interval, compensation of employees per unit of 
output rose by about 8 per cent (as wages and fringe benefits rose faster 
than productivity) . The decline in profit margins offset part of that increase 
so that unit value added rose only 3 per cent. 8 Declining raw-materials 
prices more than offset the rise in unit value added so that wholesale prices 
of finished manufactures fell by about one per cent. 

From 1955 onward the rise in compensation of employees continued. 
From 1955 to 1956 compensation of employees per unit of output rose 
another 4.7 per cent. Unit value added rose 3 per cent (profit margins 
improved but this was due to reductions in unit overheads rather than to 
a rise in the margins of price over direct costs). Wholesale prices rose 
about 3 per cent as raw-materials prices rose in line with other costs. 

During 1957 compensation of employees per unit of output rose by 
2.5 per cent and unit value added by about 2 per cent. 

Over the whole period from 1949 to 1957, compensation of employees 
rose by 26 per cent, while unit value added rose 21 per cent due to the 
reduction in profits as a percentage of value added. Wholesale prices of 
manufactured goods rose only 17 per cent due to the fact that raw- 
materials prices rose only slightly over the whole period. 

In the period from 1949 to 1951 labor costs, profit margins and raw- 
material prices were all rising. From 1951 to the end of 1954 labor costs 
were rising but raw-materials prices were falling and so were profit margins. 
From 1954 to 1955 profit margins rose, labor costs fell slightly (primarily, 
because of the improvement in the utilization of overhead labor; produc- 
tion worker payroll costs per unit rose) and raw-materials prices fell. From 
1955 to the beginning of 1957 raw-materials prices, wage costs and profit 
margins all rose. 

It seems to me that a fair conclusion from all this is that over the last 
10 years prices have risen primarily because labor costs have risen. The 
ups and downs of raw-material prices and profit margins sometimes rein- 
forced, sometimes offset the trend in labor costs but did not contribute 
much to the upward trend of prices. 



8 The drop in profit margins per dollar of sales between 1951 and 1955 was much 
smaller in percentage points than the drop in profits as a percentage of value added, 
because sales are about twice as high as value added. 

76 



The movements of raw-materials prices and profit margins are very 
similar to those experienced in the past in response to similar movements 
in demand. If anything is unique about this inflation, it is the rapid 
upward movement in wages. 

table 1. All Manufacturing Industries (except newspapers) 



PROFIT RATIOS, 1947-58 



Period 

1947 

1948 

1949 

1950 

1951 

1951 (new series) 

1952 

1953 

1954 

1955 

1956 

1953 

1st quarter 
2nd quarter 
3rd quarter 
4th quarter 

1st quarter 
2nd quarter 
3rd quarter 
4th quarter 

1st quarter 
2nd quarter 
3rd quarter 
4th quarter 

1st quarter 
2nd quarter 
3rd quarter 
4th quarter 

1st quarter 
2nd quarter 
3rd quarter 
4th quarter 

1st quarter 



1954 



1955 



1956 



1957 



1958 



Profits as Percent of Sales 



Before Tax 


After Tax 


11.0 


6.7 


11.1 


7.0 


9.3 


5.8 


12.8 


7.1 


12.2 


5.4 


11.2 


4.8 


9.2 


4.3 


9.2 


4.3 


8.4 


4.5 


10.3 


5.4 


9.7 


5.3 


10.0 


4.3 


10.4 


4.4 


9.6 


4.3 


6.7 


4.0 


8.4 


4.3 


8.9 


4.7 


8.2 


4.4 


8.1 


4.7 


9.9 


5.1 


10.6 


5.5 


10.2 


5.4 


10.3 


5.6 


10.2 


5.3 


10.3 


5.5 


9.0 


4.9 


9.3 


5.2 


9.7 


5.1 


9.4 


5.0 


8.5 


4.7 


7.6 


4.4 



6.4 



3.4 



Source: FTC-SEC Quarterly Financial Report 



77 



Manufacturing prices by type of product 

The prices of all manufactured goods were influenced by the factors 
described above. There were, however, important differences in the move- 
ments of prices of different types of manufactured products. These differ- 
ences were due to differences in the relative importance of raw materials 
and to differences in changes in labor productivity. 

From 1951 to 1956, the wholesale and retail prices of food declined as 
reductions in raw-material costs offset increases in labor costs of proc- 
essing. After 1956, food prices turned up as the rise in farm prices 
reinforced the rise in labor costs. 

The prices of other consumer non-durables also declined after 1951, but 
the decline was much smaller and came to an end in 1955. The decline 
in other non-durables was small because raw materials are less important 
than in the case of food. 

Prices of consumer durables advanced very slightly from 1951 to 1955. 
Thereafter they advanced more rapidly. Once again declining raw-material 
prices were helpful in offsetting wage cost increases in the early stage and 
reinforced it later. There is also some reason to believe that productivity 
increase in the consumer durable field was larger than in other sectors, 
particularly in the years up to 1955. 

The prices of producers durables rose steadily after 1951. The increases 
were particularly large from 1955 to 1957. At least a part of the price 
rise for producer non-durables is accounted for by the importance of steel 
in their products. Steel prices have risen more than those of most other 
products since 1951. This in turn has been due to (a) a greater-than- 
average increase in labor costs, and (b) a greater-than-average rise in 
profits per unit of output. 

Average hourly earnings in all manufacturing increased 41 per cent 
from 1950 to 1957 while hourly earnings in Iron and Steel rose by 50 per 
cent. Profit margins on sales in all manufacturing fell from 12.8 per cent 
in 1950 to 9.5 per cent in the first half of 1957. Profit margins on sales 
in Iron and Steel fell from 15.5 per cent in 1950 to 14.3 per cent in the 
first half of 1957. 

The causes of wage inflation 

The movements of raw-material prices, the special factors influencing serv- 
ice prices and the cyclical variations in profit margins confuse the picture 
of postwar inflation. But when all those influences are sorted out, it seems 
clear enough that the primary factor underlying the general rise in price 
levels since 1951 has been the steady upward movement of labor costs. 
We cannot understand the inflation unless we can understand why wages 
have risen faster than productivity. 

78 



The most spectacular wage increases are those negotiated by the large 
trade unions in industries like steel and autos. Every two or three years 
we are treated to the spectacle of negotiations between the representatives 
of the Steelworkers Union and the steel companies, or the U.A.W. and the 

table 2. Manufacturing by Primary Iron and Steel Industries 



PROFIT RATIOS, 1947-58 
(per cent) 



Year 

1947 

1948 

1949 

1950 

1951 

1951 (new series) 

1952 

1953 

1954 

1955 

1956 

1953 

1st quarter 
2nd quarter 
3rd quarter 
4th quarter 

1st quarter 
2nd quarter 
3rd quarter 
4th quarter 

1st quarter 
2nd quarter 
3rd quarter 
4th quarter 

1st quarter 
2nd quarter 
3rd quarter 
4th quarter 

1st quarter 
2nd quarter 
3rd quarter 
4th quarter 

1st quarter 



1954 



1955 



1956 



1957 



1958 



Profits as Percent of Sales 



Before Tax 


After Tax 


10.9 


6.6 


12.4 


7.4 


11.1 


6.5 


15.5 


7.9 


15.9 


5.7 


16.0 


5.8 


9.7 


4.7 


12.6 


5.3 


10.5 


5.3 


14.5 


7.2 


13.2 


6.7 


13.3 


5.2 


14.8 


5.2 


13.9 


5.4 


7.7 


5.3 


9.6 


4.7 


10.8 


5.2 


9.7 


4.6 


12.0 


6.7 


13.6 


6.6 


15.1 


7.3 


14.0 


6.9 


15.2 


7.9 


14.8 


7.3 


14.5 


7.2 


7.7 


4.1 


14.5 


7.5 


14.5 


7.1 


14.2 


7.0 


12.1 


6.1 


11.0 


5.8 



8.2 



4.2 



Source: FTC-SEC Quarterly Financial Report 



79 



auto companies gathering around the bargaining table. After much pub- 
licity, threats of a strike or an actual strike lasting a few weeks, a settlement 
is reached. In every contract since the war the steel workers, the auto 
workers and several other large unions have obtained very substantial 
wage increases. 

It is hardly surprising that many observers of these proceedings have 
concluded that the power of trade unions is responsible for the rapid rate 
of increase in wages in the past few years. That type of observation has 
led to the widely prevalent notion that we have been suffering from an 
inflation due to "cost-push" rather than to the classical type of inflation in 
which wages are pulled up by demand. 

But the situation is really much less clear than it appears when we 
consider only large-scale wage negotiations. Only about one-third of pri- 
vate non-agricultural wage and salary earners are trade union members. 
Allowance must be made for the fact that, when wage earners are organ- 
ized, the wages of non-union employees (i.e. clerical and supervisory 
personnel) of companies are strongly influenced by union wage rates. On 
the other hand a substantial proportion of union membership is in unions 
much less powerful than the steel and auto workers. 

Because of the problem created by fringe benefits and variations in 
overtime it is not easy to make precise comparisons of wage increases in 
different industries. But it is fairly clear that the wage increases obtained 
by the strongest unions in the period since the beginning of World War II 
have not exceeded the general average by any large percentage. 

Average hourly earnings in the steel industry in 1956 were just about 
triple those of 1940. The average increase in wages for all manufacturing 
was just about the same. The increase for autos was actually below the 
average for all manufacturing. Wage increases in durable goods manu- 
facturing, where unions appear to be strongest, have not exceeded those 
in the less well organized non-durables sector (in percentage terms). Data 
on wages in the service trades are far from adequate, but wages changes 
there appear to be about as large as in manufacturing. 

Some years ago there was a good deal of statistical controversy over the 
question whether wage increases in organized industries were larger than 
those in other industries. Because different answers can be obtained by 
different methods of comparison, complete agreement has not yet been 
reached. But the important conclusion is that at best union wages (plus 
fringe benefits) have not increased much more than non-union wages. 

In view of those considerations, one cannot attribute the rapid increase 
in wages in the last few years to trade union power by simply noting the 
large increases obtained by unions in large-scale negotiations. Those who 
wish to support the "cost-push" hypothesis must explain how it is that wage 
increases obtained by trade unions have been so fully reflected in unor- 
ganized industries. 

80 



On the other hand, those who wish to maintain that wages have risen 
because of demand-pull have their troubles also. During wars wages rise 
rapidly, and there is obviously a shortage of labor. It is then perfectly 
clear that wages are rising because employers find it profitable to bid more 
to get additional labor. 

In the years since 1951 we have not had large labor shortages like those 
of the war years. Nonetheless, there is evidence that there have been short- 
ages of the types of labor employers normally wish to obtain. The evidence 
for that statement is the very large increase in the employment of older 
married women, the willingness of employers to take on part-time workers, 
the prevalence of "moonlighting," and the ease with which migrants from 
farms and Puerto Rico have been absorbed into the urban labor force. 
Those classes of marginal workers can be drawn into employment only in 
conditions when there is excess demand for labor of the type usually em- 
ployed. A shortage of "bodies" has been avoided by reducing the apparent 
quality of the labor force. (I say apparent because many "marginal" work- 
ers perform just as well as "normal" workers even though employers are 
reluctant to hire them until they have had some experience with them.) 

In these circumstances one would expect that wages would rise to some 
extent. When there is a shortage of the types of labor normally hired, 
employers may be expected to compete for the normal labor force by 
raising wages before they fall back on the expedient of hiring marginal 
workers. (That is especially so when there is difficulty in making wage 
differentials correspond to employers' beliefs about quality differentials.) 

The question is whether the labor shortage or demand-pull explanation 
of wage movements can account for wage increases as large as those which 
have occurred in the last few years. We have no absolute standards by 
which to judge how much wage increase should be associated with a given 
situation in the labor market. We can only ask whether wages are now 
responding to labor market conditions in the same way as in the past 
when trade unions were less important. 

The only comparison worth making is that between the post- 1951 period 
and the years 1923-1929, and that is not entirely satisfactory. The labor 
market in the 20's was roughly similar at least to the labor market since 
1951. Average levels of unemployment were apparently about the same 
in the two periods. 

In both periods workers from farms were absorbed into the labor force; 
there was an increase in labor market participation by women in both 
cases. However, both the last two movements have been on a greater scale 
in the more recent period than in the earlier one. Moreover, there is no 
evidence of moonlighting or of general increase in employment of part- 
time workers during the 1920's. Finally, it is probable that if unemploy- 
ment were measured on the same basis in the 1920's as now, the unemploy- 

81 



ment figures for the twenties would be higher than those cited above. 
Unemployment compensation increases reported unemployment, since 
women who lose one job and do not actively seek another report them- 
selves unemployed in order to draw compensation. 

All things considered, then, it is probable that labor markets have been 
appreciably tighter on the average since 1951 than was the case during 
the 1920's. 

The movements of wages in the two periods were very different. From 
1923 to 1929 average hourly earnings in manufacturing increased from 
$.522 to $.566 or about 8 per cent. The average annual increase was 
therefore only about one per cent per year. From 1950 to 1957 average 
hourly earnings in manufacturing increased from $1,465 to $2.07 or by 
41 per cent. The average annual increase was over 5 per cent per year. 
Allowance for fringe benefits would make the difference between the two 
periods even larger. 

Admitting that labor markets have been somewhat tighter in recent years 
than they were in the 20's, it still seems very difficult to explain the differ- 
ence in wage movements by the difference in the supply and demand 
position in the labor markets. 

We have to conclude that neither the "trade union push" approach nor 
the "demand-pull" theory is really adequate to explain recent wage move- 
ments. But, after all, the two approaches are not mutually exclusive. 
There is no reason why both considerations cannot be given some weight 
in the explanation of wage movements. Moreover, as I shall argue below, 
there are strong interactions between the influence of trade unions and the 
influence of demand on wages. 

As I have already indicated, the demand for labor has been expanding 
at a faster rate than the "normal" labor force. If labor markets were 
perfect, with uniform wages for similar workers in all sectors, there would 
have been (even without trade unions) a general competition for workers 
which would have raised wages generally. But labor markets are far from 
perfect. Wages in manufacturing, construction and public utilities and 
transportation are far above those in the trade and service sectors. In 
those circumstances, a shortage of labor causes much greater recruiting 
difficulty in the trade and service sectors than in manufacturing, construc- 
tion, utilities and transportation. In the absence of strong trade unions, 
we should have expected wages to rise in the trade and service sectors. 
But because of the large initial differentials those wage increases would 
not have prevented the other sectors from absorbing the bulk of the net 
increase in the urban male labor force even if there were no wage increases 
in manufacturing, construction, utilities and transport. Wages in the trade 
and service sectors would have risen enough to make it profitable for em- 
ployers in those sectors to recruit or accept marginal workers. Of course, 

82 



a narrowing of differentials would have helped to hold workers in trade 
and service so that the other sectors would have had to raise wages some- 
what. But relatively small increases would have been sufficient. 

In fact, of course, the trade and service sectors have raised wages. Dif- 
ferentials for occupations in which women cannot readily be substituted 
for men have narrowed, and the trade and service sectors have avoided 
further wage increases by using older women, teen agers, moonlighters and 
so on. But in the other sectors, wage increases have been just as large as 
in the trade and service sectors. The large wage increases in those sectors 
were to a large extent due to the influence of trade unions. They would not 
have occurred as a result of competition for workers in the absence of 
trade unions. 

Thus the first instance of the combined influence of excess demand for 
labor and trade unions is a simple one. Excess demand for labor pulled up 
wages in the sectors in which trade unions are weakest, and trade unions 
pushed up wages in sectors where excess demand was weakest. 

But other interactions are also important. 

First, the wage increases obtained by unions are not independent of the 
balance of supply and demand for labor. When labor markets are tight, 
employers have a number of incentives for raising wages even when they 
do not suffer from a general shortage of labor. When there is little un- 
employment, there are likely to be shortages of particular types of labor. 
Employers may be willing to raise the wages of those occupational groups 
in short supply and may find it necessary to extend the increases to other 
workers to maintain the internal wage structure. 

When jobs are very easy to get, turnover rates tend to rise and employers 
may be prepared to give wage increases to reduce turnover. Similarly it 
pays employers to err on the side of generosity in disputes over piece-rate 
changes associated with changes in methods. Thus employers have a 
number of incentives to raise wages when labor markets are tight even in 
the absence of a trade union or an obvious labor shortage. 

Moreover, the bargaining position of trade unions depends on conditions 
in the labor market. In the short run the bargaining position of a trade 
union is stronger when employment is high because the financial position 
of the union and its members will then be strong enough to permit an 
effective strike. Over longer periods, high employment makes things easier 
for trade unions because it reduces the possibility of competition by non- 
union employers. 

To summarize, a very strong union may occasionally obtain a wage in- 
crease in the face of substantial unemployment, but in general the effective- 
ness of the pressure on wages exerted by trade unions will be much greater 
when there is little unemployment. 

The wage increases obtained by trade unions will also have an influence 
on wages in unorganized sectors, but the extent of that influence will 

83 



depend on the balance of supply and demand in the labor market. When 
the workers in unionized manufacturing firms obtain wage increases, 
workers in similar occupations, e.g. clerical workers and common laborers 
in the trade and service sector, will become dissatisfied if their wages are 
not increased. If jobs are easy to get, the result of such dissatisfaction will 
be an increase in turnover and a reduction of productivity. Employers in 
the trade and service sector will have an incentive to raise wages to improve 
morale and reduce turnover. 

It was argued above that neither "demand-pull" nor "trade union pres- 
sure" taken alone gives an adequate explanation of the wage inflation of 
the past few years. When we take account of their joint effects and the 
fact that demand-pull works best in one sector and trade union pressure 
in another, we do have an explanation which seems to be reasonably con- 
sistent with the facts. 



The sources of demand 

In the last section we concluded that the inflation of wages was due to an 
interaction between the pressure on wages exerted by trade unions and a 
more or less chronic shortage of labor. Either of these forces would by 
itself have caused wages to rise, but the two together caused a substantially 
greater increase in wages than the sum of those two separate effects. 

If we can explain why demand for labor was so large relative to labor 
supply and why trade unions were so effective, we shall have gone some 
distance in explaining the inflationary process. I shall not attempt to 
explain how trade unions have gotten where they are; but I shall attempt 
to outline briefly the factors responsible for the high level of demand 
which has prevailed in the postwar years. 

I shall not give a play-by-play account of the movements of demand. 
I shall confine myself to a few of the special factors which have made 
aggregate demand particularly buoyant since World War II. Three such 
factors seem especially worthy of note: (1) the position of the economy 
at the end of the war, (2) the rapid growth of population, (3) the rise 
in government expenditures. 

The situation at the end of the war 

At the end of the war private demand was unusually strong for several 
reasons. Households had large stocks of liquid assets in relation to their 
income while the ratio of consumer debt to income was unusually low. 
At the same time, stocks of automobiles and other consumer durables were 
unusually low in relation to income. As a result, households spent an 

84 



unusually high proportion of their income. Consumer saving during 1946, 
1947 and 1948 was only about half its normal level. 

Businesses found themselves in much the same position with a large 
amount of liquid assets, a low level of indebtedness and a deteriorated 
stock of physical capital. Demand for capital goods was therefore strong 
at the end of the war. 

The rise in income during the war, the strong financial position of house- 
holds and the large number of recent marriages provided the basis for a 
residential construction boom which was slowed down only by the limited 
availability of building materials and skilled manpower. 

Finally, state and local governments were faced with demands for all 
sorts of public improvements and were in a strong financial position so that 
they were both willing and able to borrow. 

Population growth 

The large number of marriages and births in the years after the war had 
two important effects. First, it was a major factor in the postwar housing 
boom. This boom, though initially based on the backlog of demand 
resulting from low rates of construction during depression and war, could 
not have continued at so high a level for so long a period without the 
support provided by high marriage and birth rates. The rapid growth of 
housing construction in the years between the end of World War II and 
1950 was one of the major forces behind the growth of demand during 
that period. In addition to its influence on housing, the growth of popu- 
lation has raised the level of consumption relative to income. It is not easy 
to measure this effect, but some rough estimates indicate that the growth 
of population has raised consumer expenditures by from $1 to $2 billion 
per year in each year since the war (over what they would have been 
with constant population). 

Effect of government expenditures 

The initial situation at the end of the war reinforced by the rapid growth 
of population was adequate to provide the basis for a strong postwar boom. 
The boom was interrupted by the recession of 1949, but it can reasonably 
be assumed that had the Korean War not intervened there would have 
been a strong recovery. By the end of 1949, recovery was already under- 
way as residential construction increased; federal expenditures rose from 
the low levels (which had permitted large surpluses in 1947 and 1948) 
while tax rates had fallen, and business sentiment recovered from fears of a 
great postwar depression. 

Without the Korean War and the subsequent expansion of federal ex- 
penditures, income would undoubtedly have expanded after 1949. The 

85 



rate of growth of real demand would, however, have been substantially 
lower than was actually the case. We cannot calculate exactly the effect 
of the rise in government expenditures, but we can get a rough idea of its 
order of magnitude. 

During the period from the beginning of 1950 when the federal budget 
was roughly balanced until the end of 1953, federal government purchases 
of goods and services expanded by about 40 billion dollars in 1958 prices, 
and the deficit increased by about 9 billion dollars. At a rough estimate, 
the net effect of those changes was to increase aggregate demand by 54 
billion dollars in 1958 prices or by about one half of the total increase over 
the period. That figure was obtained by applying a multiplier of 2.6 to the 
increase in the deficit and a multiplier of 1 to the balance of 31 billion of 
increased expenditures financed through taxes and assuming that private 
investment was unaffected by the government's contribution to the rate of 
increase of income. Those figures are of course significant only in terms 
of their order of magnitude. 

The actual difference in Gross National Product (GNP) would have 
been still greater. For had the rise in GNP been slower than it actually 
was, private investment would have been lower which would have further 
retarded the rise in GNP. That reduction in the growth rate would have 
had still further secondary effects and so on. 9 

Some of those secondary effects were not felt, however, until after 1953. 
Housing construction and state and local building were held down during 
the Korean War period, and at some points plant and equipment expendi- 
tures were retarded by shortages. During the period from 1950 to the 
end of 1953, then, it can be said that the federal government was respon- 
sible for at least half and probably substantially more than half of the 
increase in aggregate real demand. 

After 1953, the federal government's fiscal policy worked in the opposite 
direction. Federal expenditures declined about 20 billions (in 1958 prices) 
between the end of 1953 and the end of 1957, while the deficit of 9 billion 
in 1953 was turned into a two billion dollar surplus. The effect of those 
changes was to reduce income (by comparison with what it would have 
been with constant expenditures and deficit) by about 37 billions. 



9 The conclusion reached above as to the magnitude of the effect of a balanced 
budget reduction in expenditures before taking account of the effect of the difference 
in GNP is probably conservative. Tax reductions for business instead of consumers 
would lead to a decline in consumption and some increase in investment by com- 
parison with the case in which all the tax benefits go to households. It is hardly 
likely that the average propensity to invest out of profits after taxes can be as high 
as the average propensity to consume out of disposable income. 

Any more realistic assumption about the distribution of tax benefits would imply 
an income redistribution in favor of high savers and increase the adverse effect of 
expenditure reduction. 

86 



The restrictive effects of federal policy were offset by the rise in private 
investment and the rise in state and local expenditures. The high volume 
of private investment and the rapid growth in state and local expenditures 
after 1953 were at least in part due to the rapid growth of income before 
1953 and to the restriction of construction during the Korean period. 

The restriction of federal expenditures after 1953 slowed down the rate 
of growth of income in drastic fashion. From the cyclical peak of 1953 to 
the peak in 1957, GNP in real terms increased by about 2.5 per cent per 
year which was less than half the growth rate achieved between early 1950 
and the end of 1953. 

Unfortunately, the rate of growth of productivity was also somewhat 
lower than in earlier years and the male labor force continued to grow at a 
very slow rate. The reduction in government expenditures therefore did 
little more than to keep the growth of demand for labor in line with the 
growth of supply without reducing the tightness in labor markets which 
had existed in 1953. 

Over the whole period, the primary effects of federal fiscal policy were 
relatively small, but because of their timing the period since Korea reflects 
most of the secondary effects of the expansion of government expenditures 
prior to 1953 and only a part of the secondary effects of the restriction 
since 1953. Over the whole period since the Korean War, therefore, fed- 
eral fiscal policy made a substantial net contribution to the generation of an 
excessively high rate of growth of demand for labor. 



Can we control inflation by controlling demand? 

Those people who believe that inflation is caused by trade union pressure 
on wages have frequently maintained that we cannot maintain a satisfactory 
level of employment and control inflation by controlling demand. They 
take the view that in order to prevent wage inflation it is necessary to have 
very large amounts of unemployment. They conclude that we must either 
let the inflation proceed or control wages and prices by some other method 
than by controlling demand, i.e. some kind of direct control of wages and 
prices or by some action to reduce the power of trade unions. Their posi- 
tion has been strengthened by the fact that the recessions of 1953-54 and 
1958 slowed down but did not stop the rise in wages. 

If it were true that all workers were organized in powerful trade unions 
and that all the inflation of wages were due to trade union pressure, their 
conclusion might well be correct. However, that is not the case. There has 
been a good deal of excess demand for labor in some sectors of our econ- 
omy in the last few years and that has caused wage increases in those 
sectors directly and has reinforced the effects of wage increases negotiated 
by trade unions elsewhere. That conclusion suggests that the average rate 

87 



of increase in wage rates might be considerably reduced if we could bring 
the average rate of growth of demand for labor into line with the average 
rate of growth of the normal labor force over a period of several years. 

Suppose that were done. The consequence would be, I think, a sub- 
stantial reduction in the rate of increase of wages in the unorganized 
sectors of the economy. There would still be wage increases in all sectors. 
Trade unions would still be able to negotiate wage increases, and those 
increases would still have some effect on wages of unorganized workers. 
Over a longer period a better balance between growth of demand for labor 
and growth of normal labor supply would tend to reduce the effectiveness 
of trade union demands for wage increases. For if wages rose slowly in the 
unorganized sector and rapidly in the strongly organized ones, the differen- 
tials between the two would widen. It does not seem likely to me that 
differentials between union and non-union wages can be widened indefinite- 
ly, particularly when labor markets are not too tight. A widening of 
differentials would encourage the entry of non-union employers and in- 
crease the effectiveness of the competition of products produced by unor- 
ganized workers. (Unfortunately, of course, there is the possibility that a 
widening of differentials would lead to extension of trade union organiza- 
tion into new areas of the labor market.) 

A policy of holding down the rate of growth of demand would be costly 
in a number of ways. Many people believe that we need a high rate of 
growth of aggregate income (to compete with the Russians, to make 
possible an increase in public expenditures for defense, education, etc.). 

Secondly, a high-pressure economy in which capacity and labor force are 
fully utilized may be likely to have a higher rate of growth of productivity 
than one in which there is less demand pressure. Full utilization of capacity 
increases the prospective profits from the installation of improved equip- 
ment and also raises the level of current profits which makes it easier to 
finance these improvements. The existence of tight labor markets increases 
the profitability of developing and installing labor saving devices. Against 
those effects must be set the reduction of efficiency on the part of both 
labor and management which results from high profit margins and tight 
labor markets. 

Finally, a reduction in the rate of growth of demand for labor is likely 
to result in an increase in the level of unemployment. If the rate of growth 
of aggregate demand for labor is about in balance with the rate of growth 
of the normal labor supply, there are bound to be areas in which the rate 
of growth of demand for labor is less than the rate of growth of normal 
labor supply. In addition, there will be absolute reductions in employment 
in some industries. Since workers do not transfer instantly from one indus- 
try or area to another, there is bound to be greater unemployment in the 

88 



normal labor force than there would be with a higher rate of growth 
of demand. 



Conclusion 

We have argued that the inflation is due to both demand-pull and trade 
union pressure on wages. A reduction in the rate of growth of the demand 
for labor would reduce the rate of increase of wages even in the face of 
the wage pressure exerted by trade unions. Some of the resulting gain 
might, however, be offset by a reduction in the rate of increase of produc- 
tivity. In addition, a reduction in the rate of growth of demand would be 
costly in itself and would lead to higher levels of unemployment. 



89 



4 




• The impacts of unions 
on the level of wages 



Clark Kerr 



Inflation and industrialization have marched together for the past two 
centuries. Rising prices and growing industrial output have characterized 
much of the economic history of this period, with the notable exception of 
the second half of the nineteenth century. While they have marched to- 
gether, they have not marched closely and evenly in step until the past few 
years. The alternations of war and peace and prosperity and depression 
have variously affected the course of both inflation and industrialization. 
The pace of both has been subject to great variations. 



Clark Kerr is President of the University of California, Professor of Business 
Administration and Economics, and Research Associate in the Institute of Industrial 
Relations, Berkeley. He was Chancellor of the University's Berkeley campus from 
1952 to 1958. Previously he had taught, and served as Director of the Institute of 
Industrial Relations at Berkeley (1945-52). Additionally, he has served as Consultant 
to the Departments of State (1950) and Labor (1954), as a member of the 6th and 
12th Regional War Labor Boards during World War II, and on the National Wage 
Stabilization Board as public member and vice-chairman (1950-51). He has acted 
as arbitrator in numerous labor disputes, and presently is a Director of the Center 
for Advanced Study in the Behavioral Sciences of the Ford Foundation. Dr. Kerr is 
the author of articles in various scholarly and professional journals, and has con- 
tributed sections to numerous books on industrial relations and the social sciences. 
Dr. Kerr acknowledges the assistance of Margery Galenson in the preparation of 
this paper. 

91 



The past few years, since World War II, on the other hand, have been 
marked in Western Europe and North America by more nearly constant, 
rather than sporadic, inflation and industrialization. And, while a little less 
than a decade and a half constitutes a great deal less than an everlasting 
trend, a spectre is haunting Western capitalism — the spectre of constant 
inflation. Many, although not all, of the powers of society have been allied 
to exorcise this spectre — government officials and editorial writers, mone- 
tary authorities and economists, financiers and ministers — but it still exists. 

It is a real spectre though it is not new and it need not make society 
tremble. It is a real spectre because constant inflation could become con- 
stantly greater inflation; because it redistributes income often in an inequi- 
table fashion; and because policies to combat it may also combat progress. 

Constant inflation presumably should have a constant source. Several 
new or relatively new developments have accompanied constant inflation 
and are, consequently, the most likely causes. One of these is the growth 
of the trade union movement and of its power over the wage-setting 
process. But it is not the only companion. Other companions have been 
governmental commitments to full employment, policies and practices 
leading to unbalanced budgets and low interest rates, rapid industrial ex- 
pansion, new supply conditions in labor markets and new patterns of 
mobility and immobility, and new mechanisms for price control by private 
agencies; and several of these companions are closely related to each other. 

The question here is the responsibility of one of these companions — 
the trade union — for constant inflation. This is not an easy question to 
answer, partly because of the intermingling of ideology and group self- 
interest with analysis in so much of the discussion, but particularly because 
with so many things happening it is almost impossible to state precisely 
the force of any one development by itself. For example, what would be 
the effect of the trade union if there were no industrial growth or if there 
were no administered prices to go along with administered wages? Conse- 
quently, reliance must be placed on individual judgments rather than any 
universally accepted analysis; and judgments have differed. 

The split jury 

The jury which has sat most constantiy on this case has been composed 
of economists; and almost any conceivable verdict can be obtained by 
picking almost any conceivable economist. 1 To illustrate: 

To Lindblom 2 the union is a "monopoly" and also a "body politic." As 
a body politic, under the urging of political pressures, it uses its monopoly 



1 For a summary of the recent literature see G. H. Hildebrand, "The Economic 
Effects of Unionism," in A Decade of Industrial Relations Research, Harper & Bros., 
1958. 

2 Charles E. Lindblom, Unions and Capitalism, Yale University Press, 1949. 

92 



power to force wages higher and higher. This leads to "unemployment or 
inflation" and, with government guarantee of full employment, to inflation. 
As a result, "unionism and the private enterprise system are incompatible." 

To Chamberlain 3 the unions introduce a "monopoly element" into the 
labor market and, whether or not they try to maximize the wage bill, they 
do try to get "more" and this leads to "wage-push inflation." "Unions 
today do have too much economic power." 

To The Economist 4 the real cost of trade unions is not so much the loss 
in productivity per man-hour they cause but rather that they turn full pro- 
duction into full inflation; and to avoid the latter, the former must also be 
forgone. This is one of the great economic tragedies of our age and our 
type of society. 

To Hicks 5 the "Labour Standard" has replaced the Gold Standard. 
Governments will adjust their policies to maintain full employment at what- 
ever wage levels the unions choose to set; and price levels follow along. 
But the unions, or at least British unions, may not be so unreasonable that 
this "Labour Standard" is much more "dangerous" than other monetary 
systems. 

To Lerner 6 the problem is not "wage-cost inflation" alone but "seller's 
inflation." For there is also "profit inflation" as well as "wage inflation," 
and it is very difficult and even impossible to untangle the two. Wherever 
there are administered prices and administered wages, and they seem to be 
nearly everywhere, "seller's inflation" is a possibility, and it must be dealt 
with as a unitary phenomenon. 

To Slichter 7 the unions are only one of several causes of inflation, and 
the others include the reduced availability of new sources of labor and the 
policy of government; but they are a significant cause. He concludes that, 
between 1933 and 1953, unions pushed up the general wage level "at least 
25 cents per hour and probably more." This is one-fifth of the total in- 
crease that occurred during that period. 

To Reynolds* "collective bargaining does not have as much impact on 
the money-wage level as has sometimes been suggested. My judgment 
would be that between 1945 and 1955 the money-wage level rose little, 
if any, more than it would have risen under nonunion conditions." 



3 Edward H. Chamberlain, "The Economic Analysis of Labor Union Power," in 
Labor Unions and Public Policy, American Enterprise Association, 1958. 

4 August 2, 1958. 

5 J. R. Hicks, "Economic Foundations of Wage Policy," Economic Journal, Septem- 
ber 1955. 

6 A. P. Lerner, "Inflationary Depression and the Regulation of Administered Prices," 
United States Joint Economic Committee, The Relationship of Prices to Economic 
Stability and Growth, March 21, 1958. 

7 Sumner H. Slichter, "Big Unions and Inflation," American Economic Review Pro- 
ceedings, May 1954. 

8 Lloyd G. Reynolds in New Concepts in Wage Determination, McGraw-Hill, 1957. 

93 



To Morton 9 unions are a minor factor affecting inflation and may retard 
it as well as augment it: retard it in a boom period; increase it in certain 
industries, where government regulation relates prices to costs, like public 
utilities and railroads. 

To Friedman 10 unions have both a "rigidity effect" and "upward-press- 
ing effect." The former holds down wage levels in a period of expansion; 
the latter forces them up in a period of stability. The two largely offset 
each other; but, of the two, the rigidity effect may be the more important 
under recent circumstances. 

To Boulding u it is a certainty "that the main effect of unionism is to 
hold down wages and to prevent them from rising faster than they other- 
wise would . . . Unions are the opiate of the people under capitalism. 
That is why you have got to have them." 

From the destroyer of "private enterprise" to the "opiate of the people," 
from the source of disastrous inflation to a bulwark of price stability, from 
a powerful monopoly to a minor or even negative force — the judgments 
vary. Economics is not yet a science; but economists are certainly free 
thinkers. 

As a very part-time economist, I should like to suggest that all of them 
are right and all of them are wrong. All of them are right to the extent 
that they suggest that some kinds of unions could have the suggested effects 
under some kinds of circumstances. All of them are wrong, to the extent 
they suggest (and some of them do not) that their conclusions are the 
universal rule. The only universal rule is that there are all kinds of unions 
operating under all kinds of circumstances and they can have all kinds of 
effects. But it should also be added that kinds and circumstances and 
effects can be related — at least to a certain degree. Truth is more likely 
to emerge from studying the impacts of the unions, than "the impact of 
the union." 

Types, circumstances and impacts 

Types 

When talking about unions, it is helpful to specify the kind of union one 
is talking about. In terms of their approaches to price stability, unions can 
be broadly divided into the following general types: 

Agent of the State — The "agent of the state" union, as in Russia or 
China, is the willing tool of the national administration. It serves its 



9 Walter A. Morton, "Trade Unionism, Full Employment and Inflation," American 
Economic Review, March 1950. 

10 Milton Friedman in Impact of the Union, Harcourt Brace & Company, 1951. 

11 Kenneth E. Boulding, idem, p. 245. 

94 



policies. It has no policies of its own. It is a weapon of social discipline, 
and the only variation of which it is capable is in the degree of its effec- 
tiveness. 

Partner in Social Control — Some unions serve as "partners in social 
control." They may be formal partners, as they have been in Holland, 
assuming joint public responsibility for the economic welfare of the nation; 
or they may be informal partners, as they have been in Germany, almost 
equally committed with the government to reasonably full employment and 
reasonable price stability at the same time. In Britain and in the Scandi- 
navian countries, the unions have served as such informal partners when 
Labor or Social Democratic parties were in power. Here again there can 
be degrees of effectiveness as "social partner." 

Sectional Bargainer — The union, as "sectional bargainer," is concerned 
not with the national impacts of its actions but with the consequences for 
its members and for its industry or segment of an industry. Its responsibil- 
ity is relatively narrowly denned. The United States and Canada are 
representative of this type of unionism. The "sectional bargainer" union 
may be found in two major phases — (a) a state of excitement and (b) a 
state of normality. A state of excitement is most likely to exist in a new 
union, a union subject to the challenge of a rival union or a union under- 
going internal political upheaval; and bargaining is likely to be much more 
aggressive in a state of excitement than in a state of normality. 

Class Bargainer — The union, as "class bargainer," endeavors to get a 
"fair share," which usually means a larger share, of the national income for 
labor as a whole. It is usually matched by other "class bargainers," as in 
France, who seek "fair shares" for agricultural producers, the commercial 
classes, the civil servants, and so forth; and the total of these "shares" is 
almost certain to add up to more than the national output of goods and 
services. The "class bargainer" union usually has or develops a class 
ideology. 

Enemy of the System — The "enemy of the system" union is devoted to 
the destruction of the surrounding economic and political structure. Among 
its techniques are the sabotage of production and the encouragement of 
excessive consumption aspirations. Such unions have been really effective 
only when a society is in the process of disintegration. 

These above types suggest more uniformity and stability than is the 
actuality. Some societies have mixtures at any one moment of time — as in 
France with Communist, Socialist and Catholic unions. In some societies, 
the union movement shifts from one "type" or policy to another. The 
"agent of the state" union will remain an "agent of the state" so long as 
the state needs an agent. But the "social partner" union may be a partner 
only when the nation faces an emergency or when a government it favors 

95 



is in power and then turn to a "sectional" or "class" approach under other 
circumstances. The "enemy of the system" union may in non-revolutionary 
periods follow a "class bargainer" policy instead of open full-scale opposi- 
tion, or even be a particularly belligerent "sectional bargainer." 

Each of these types has its own most natural habitat — the "agent of the 
state" union in an authoritarian society; the "social partner" union in a 
"social democratic" context; the "sectional bargainer" in a free enterprise 
system; the "class bargainer" in a semi-class or semi-feudal society; and 
the "enemy of the system" union in the latter type of society in the course 
of its decay. A society does not just conjure up the kind of unionism it 
would like to have after looking at the different models theoretically 
available; some kinds fit some societies and not others. 

But we are concerned here not with the ultimate cause of a certain type 
of unionism, but with its impact on inflation. In general, unions — if they 
may all truly be called unions — will make a contribution to economic 
stability in the following descending order: 

Agent of the state 

Partner in social control 

Sectional bargainer 

Class bargainer 

Enemy of the system 
The merest glance at this list indicates that a society usually cannot pick 
its type of unionism on the basis alone of its impact on stability; and that 
its effect on the price level cannot be the only proper test of the desir- 
ability of a union movement. 

Theoretically, however, it might be expected that unions, from the top of 
the list to the bottom, would vary from strong supporters of stability to 
effective agents of instability. 

Circumstances and impacts 

Unions, of whatever type, operate within an environment, and their 
potential impacts on the general level of money wages may be almost as 
much related to the environment as to their type. 12 Among the environ- 
mental situations with which we shall treat are those relating to the policies 
of other institutions (government and employers), to employment condi- 
tions, and to labor market conditions. 



1 2 We are discussing here only the impact on the general level of money wages and 
not the impact on prices. That depends also on changes in the level of productivity 
and in labor's share of income. Also we are discussing the economic impact of the 
unions within a given environment and not their political or economic impact in 
their efforts to change the surrounding environment. 

96 



The standard for comparison will be "what would otherwise have hap- 
pened" had there been no union; and this nobody really knows. The 
standard will not be the absolute increase in money wages; for unions may 
sometimes do most when they seem to do the least, and do least when 
they seem to do the most. For example, in a depression a union may hold 
up wages which would otherwise go down and we can say they "raised 
the level"; while in a boom period they may belatedly negotiate a sub- 
stantial wage increase which would have come earlier under non-union 
conditions through the operation of market forces, and we can say they 
"reduced the level." 

We shall consider first the policies of other institutions. Guaranteed full 
employment places the unions in an advantageous position, and two types 
of unions — "enemy of the system" and "class bargainer" — are in a par- 
ticularly good position to take advantage of it. Administered prices by 
employers create a special opportunity for the "sectional bargainer" union, 
for administered wages can be passed on through administered prices and 
turn up in administered inflation. 13 With pattern bargaining, high settle- 
ments in an area of "administered prices" are likely to be imitated in other 
areas and thus spread the high "key" settlement. When the government is 
fearful of strikes and enters the collective bargaining arena to settle dis- 
putes, this again creates a favorable environmental situation for each of the 
three types of unions just mentioned. However, were the government to 
undertake a critical public review of wage settlements, this would have the 
opposite effect, and the "sectional bargainer" and particularly the "social 
partner" unions would be sensitive to such review. Government wage 
controls create an unfavorable condition for union impact on the general 
level of money wages and especially for the "social partner" union; the 
"agent of the state" union is, of course, always subject to wage controls. 

In terms of employment conditions, unions probably have the greatest 
upward impact on money wages in a depression, when their attachment to 
past levels and the lags inherent in collectively bargained wages work to- 
ward stability. Next, in the downswing, particularly the early phases, they 
may not only be able to hold wage levels but actually increase them, con- 
trary to "normal" tendencies. In a period of stable full employment, union 
pressure may well keep wages rising at some "standard" rate, say five per 
cent a year, when under other circumstances they would have risen more 
slowly. In an upswing, particularly its later stages, and in "overly full" 
employment, however, unions with their term agreements and formal ap- 
proaches may cause a lag behind the adjustments which would otherwise 



13 Also administered prices can create profit margins which lure the unions to make 
higher wage demands. 

97 



occur. A general rule might be: the smaller the wage adjustment, the 
greater the true impact of the union; and the greater the wage increase, 
the lesser the real impact. 

Labor market conditions may also relate to union impact. In a period 
of rapid accessions to the labor force — women, migrants from rural areas, 
young people — the unions can protect wages from the depressing effects. 
But when a labor force has become immobile, due to pensions or seniority 
rules or excessively specialized training or for other reasons, the union may 
reduce the upward impact on wages of this immobility. In the absence of 
unions, employers would tend to respond to individual scarcity situations 
with selective adjustments; and the impacts of these would spread. Unions, 
with their more formal wage relationships, tend to dampen this tendency 
and force employers to make other adjustments than the bidding up of 
individual classes of skills. This may possibly serve to lower, somewhat, 
the general level of money wages. 

Putting together the variety of types of unions and the variety of environ- 
mental settings results in a variety of potential effects. Unions raise the 
general level of money wages greatly; or perhaps only a little. Unions 
reduce the general level of money wages substantially; or perhaps only a 
little. Or perhaps they have no effect at all. It all depends. And it all 
depends on type and circumstance, as the summary table suggests (see 
Table 1). 

table 1. Factors Relating to Union Impact on General 
Level of Money Wages 







Policies of Other 


Employment 


Labor Market 




Type of Union 


Institutions 


Conditions 


Conditions 


1 


^ "Enemy of the 


"Guaranteed" full 


Depression 




Raise level 


system" 
"Class bargainer" 


employment 


Downswing 


New recruits 






Administered prices 






"Sectional 




Stable full 




(As com- 


bargainer" 


Government 


employment 




pared 




settlements to 






with what 


a. State of 


avoid strikes 






would 


excitement 








otherwise 


b. State of 




Upswing 




prevail) 


normality 


Government 










review of wage 


"Overly 


Immobile 




"Partners in 


settlements 


full" 


labor 




social control" 




employ- 


force 


Reduce 




Government 


ment 




level 

> 


"Agent of the 
f state" 


wage control 







98 



The variety of experience 

Experience is different from experiment. There have been no conscious 
experiments, and in the nature of the case there cannot be, through which 
a determination could be made with accuracy of the impact of the union on 
the general level of money wages. There is only experience; and the know- 
able reality from this experience is little more than conjecture. To speak 
with full assurance in this area is to speak from prejudice or from ignorance 
or both. Yet some things can be said. 

Possible tests 

There are at least four ways in which one might try to test the impact 
of the union. 

1. How have union wages risen as compared to non-union? One might 
find here the true impact of the union not only on inter-industry and inter- 
occupational differentials but also on the general level of money-wages. 

But union and non-union wages are not in water-tight compartments and 
what happens to one set of wages may affect the other. If it were found 
that union wages went up only as fast as non-union wages, this might mean 
the unions had no impact; however, it might only mean that non-union 
wages were playing "follow-the-leader" and thus that the unions were 
having an even greater effect on the general level of money wages. Also, 
if union wages were found to be going up faster, this might imply the 
unions did have an impact; but it might only reflect the fact that the wages 
of unionized manual workers, under the impact of broadly available edu- 
cational opportunities and the breakdown of class lines, were rising faster 
than those of non-unionized white collar workers who had come into rela- 
tively greater supply — the important comparison might be manual and 
non-manual, not union and non-union. 

2. How has recent history, when strong unions existed, compared with 
earlier history when there were fewer unions? Here again it might be 
discovered how the introduction of unionism has affected the course of the 
general level of money wages. 

But the statistics, on any really comparable basis, do not go very far 
back. And if they did, it would still be true that more has happened in the 
course of intervening events than the rise of a union movement. Even ad- 
justing for the amount of unemployment, there is still the question of what 
effect the expectation of generally lower rates of unemployment would have 
had on the behavior of employers in any event. Also, since employers, 
whenever they can, tend to share their profits one way or another with 
their workers, what would have been the effect of administered prices even 
without administered wages? And what has been the consequence of the 

99 



drying up of the old sources of cheap labor on the general level of 
money wages? 

3. How has labor's share of national income behaved? If there is 
evidence that the unions have really "squeezed" profits below their "nor- 
mal" levels, then it might be said the unions were pushing wages up against 
profits and thus against prices. 

But labor's share is one of the mysteries of economic analysis. And it is 
also affected by other developments than union pressure alone. There may 
be implications to be drawn from the analysis but little or no proof. 

4. How has experience varied from one country to another? If one 
country has had a different course of money-wage levels from another, this 
different course might be related to the presence or absence of unions, or 
the different types of unionism; and we might find our answer. 

But each country varies from the other in more ways than the presence 
or absence of unions, or the nature of union policy. Also, each type of 
union policy, as we have noted, is so related to its surrounding environment 
that it is difficult to say what is the real cause of a different behavior of 
money wages — the type of union or the type of economy. 

With all their imperfections, these are four possible tests and their 
application to the actual course of events should give us some indications 
of how much and under what circumstances unions have had an impact 
on the general level of money wages. 

Actual tests 

The application of actual or presumed facts to our problem is fraught 
with a number of perils, some of which have been mentioned earlier. How- 
ever, their application may indicate a reasonable range of answers to 
our questions. 

The United States — ( 1 ) The various studies which have been made of 
the course of union and non-union wages offer no clear conclusions. Their 
results depend, to a substantial extent, on the dates taken for the studies 
and the definitions used. It may be fair to conclude, nevertheless, that, 
except for periods of active new unionism (as 1936-1937) and for situ- 
ations with a closed shop (building trades), there is little evidence of a 
definite upward push by unions on wages. 14 

(2) The history of wage movements in the United States provides some 
additional evidence. Real compensation per man-hour dropped less from 
1931 to 1932 (less than 2 per cent) when unions had strong influence in a 



14 For summaries and comment on the literature see L. G. Reynolds, "The Impact 
of Collective Bargaining on the Wage Structure in the United States," and Clark 
Kerr, "Wage Relationships — The Comparative Impact of Market and Power Forces," 
in John T. Dunlop (ed.) The Theory of Wage Determination, Macmillan, 1957. 

100 



few industries than from 1893 to 1894 (3 per cent) or 1920 to 1921 
(3 per cent). 15 Compensation in the 1931 to 1932 period held steadier, as 
compared with consumer prices, than in the two earlier periods, possibly, 
in part, because of union influence. 16 

Money wages held much steadier in 1944 to 1945 than in 1917 to 1918. 
In 1944 to 1945, wage controls were in effect by government as against 
1917 to 1918, when there was great freedom in wage adjustments. But it 
should also be noted that the unions in 1944 to 1945 accepted and even 
cooperated in the imposition of wage controls, and also that the contractual 
mechanisms which had grown up since 1917 to 1918 helped make it 
possible to exercise control over the great mass of wage rates that comprise 
our national wage structure. 

In 1936 to 1937, with new and rival unionism, money wages and real 
wages jumped much more rapidly than one would normally expect in a 
period marked with as much unemployment as then existed. 

Taking two longer periods, 1900 to 1910 and 1947 to 1957, both eras 
of quite sustained growth, it is noticeable that money wages rose faster than 
productivity in both periods. From 1900 to 1910, wages rose by one-third 
and productivity by one-fourth; from 1947 to 1957, by one-half and by 
one-third. It would appear that there may be an inflationary tendency, 
with wages rising faster than productivity, in a period of sustained growth 
under both largely non-union and largely union conditions. However, the 
excess gains of wages over productivity were somewhat greater in the 
second period and this may be due, in part, to unionism. Wages rose 
roughly one-third faster than productivity in the earlier period and one-half 
faster in the later period. 

At a productivity rate of increase of 2.5 per cent a year and assuming 
that price rises reflect the comparative changes in wages and productivity 
(in other words, that there is no change in labor's share of national in- 
come), the price impact of the greater comparative wage increase would 
be about one-half of one per cent a year. 17 But it should be remembered 
that in the period 1947 to 1957, as compared with 1900 to 1910, there 
was much less of a labor reservoir of foreign immigrants, rural migrants 
and women, that administered prices were more widely prevalent, that 
government had created the expectation of continuing full employment and 



15 For the basic statistics for this and the immediately succeeding comments, see 
Table 1 in the paper by Albert Rees in this volume. 

16 For other evidence on the increasing rigidity of wages in business contractions, 
see D. Creamer, "Behavior of Wage Rates During Business Cycles," The Relationship 
of Prices to Economic Stability and Growth, op. cit. 

17 2.5 plus one-third equals 3.33; 2.5 plus one-half equals 3.75; the difference between 
3.33 and 3.75 is 0.42. 

101 



thus less risk for the employer who raised wage rates, and that there was 
the Korean War. Consequently, unionism, by itself, cannot be held respon- 
sible for the full one-half of one per cent a year. 

Several years ago Garbarino concluded, on the basis of a study of the 
period 1899 to 1929, that, under non-union conditions, money wages and 
productivity kept pace with each other with unemployment rates of around 
5 to 6 per cent. 18 Most recent experience with unemployment rates above 
5 per cent has shown money-wage rates rising faster than long-term pro- 
ductivity rates. In 1958, with unemployment at nearly 7 per cent, hourly 
rates in manufacturing went up about 3.5 per cent over 1957, as against 
the 2.5 per cent which might be considered "normal" (the long-term rate 
of increase in productivity), but then productivity seems to have risen 
faster than normal also. However it should be noted that wage rates rose 
only about two-thirds as fast in the second half of 1958 as in the second 
half of 1957. 

(3) Labor's share of national income has tended to be quite constant 
in the long run after adjusting for changes in the proportion of wage earners 
and in the inter-industry mix. But there have been occasions when the 
profit share has been "squeezed" and the wage share increased arid, per- 
haps, partly due to union pressure on wages. These have been periods of 
depression (1931 to 1934 and 1938), periods when prices were held by 
price controls or the slower movement of administered prices in an infla- 
tionary period (1944 to 1947) and, most interestingly, a period of 
sustained full employment without substantial inflation, as in 1954-1957. 19 

Perhaps it could then be said that wages were really "pushing" on profits 
and thus on prices under these three circumstances. When there is no 
change in the profit share, it is harder to say who or what is "pushing" or 
"pulling"; and, when the profit share is rising, it would seem to indicate a 
"pull" rather than a "push." "Wage inflation," or wage pressure on the 
price level without inflation, would seem most likely to have occurred when 
labor's share had risen above "normal." 

Great Britain — ( 1 ) A recent study in Great Britain, by Phillips, 20 
relating wage increases to volume of unemployment, as Garbarino has done 



18 J. W. Garbarino, "Unionism and the General Wage Level," American Economic 
Review, December 1950. For comment on the relationship of changes in employ- 
ment and changes in wages, 1947-57, see O. Eckstein, "Inflation, the Wage-Price 
Spiral and Economic Growth," the Relationship of Prices to Economic Stability and 
Growth, op. cit. 

19 For a review of the literature see Clark Kerr, "Labor's Income Share and the 
Labor Movement," in George W. Taylor and Frank C. Pierson (editors) New Con- 
cepts in Wage Determination, op. cit.; and for statistics more recent than included in 
this review see Survey of Current Business, February 1958. See also R. and N. Rug- 
gles, "Prices, Costs, Demand and Output in the United States," The Relationship of 
Prices to Economic Stability and Growth, op. cit. 

20 A. W. Phillips, "Money Wages and Unemployment in the United Kingdom," 
Economica, November 1958. 

102 



for the United States, shows some interesting parallels and variations. 
Working with three periods, 1861 to 1913, 1913 to 1948, and 1948 to 
1957, the first marked by relatively weak and the latter two by relatively 
strong unionization, Phillips found a very close correspondence between 
the related behavior of money wages and unemployment. The 1913 to 
1948 period particularly followed the expectations based on the 1861 to 
1913 period with only one major exception. Money wages went up faster 
in the years 1935 to 1937, a time of active union revival after the depres- 
sion and also of rising food prices, than the general relation of money wage 
changes to the volume of unemployment would suggest. Taking the years 
1948 to 1957, the "wage restraint" period of Trades Union Congress policy 
showed a lower than "normal" wage advance, but the years immediately 
following the end of the policy were noted for an unusually rapid increase, 
although they were also years of a rapid rise in import prices. As compared 
with the United States, lower levels of unemployment were found to be 
associated with the same wage behavior. For example, wages and pro- 
ductivity have seemed to march hand in hand with 2.5 per cent unem- 
ployment in Great Britain rather than 5 per cent in the United States. 

(2) British experience also shows the importance of the divergence 
between actual rates and nominal rates — the rates paid in fact and those 
provided for by collective agreements and other formal documents. This 
divergence, or "wage drift," varies from one situation to another, but it was 
particularly great during periods of wage restraint, World War II and 1948 
to 1950. 21 Actual rates drifted away from control through local action of 
employers and unions. 

(3) The history of labor's share in Great Britain suggests no different 
general conclusions than does the history for the United States. Wages 
have squeezed profits when product markets were "hard" but not when 
they were "soft." 22 

Western Europe — A review of postwar experience in selected Western 
European countries, including for the sake of comparisons the United 
Kingdom and the United States, is instructive (see Table 2). France, with 
its "class bargainer" approach and the type of economy associated with it, 
has witnessed the greatest increase in the general level of money wages in 
manufacturing. Italy, however, with a somewhat similar approach, has had 
a relatively small increase. This emphasizes the point that other things 
are happening to an economy aside from union action. In Italy, over this 
period, unemployment has averaged 9 per cent, while it has been at quite 



21 See H. A. Turner, "Wages, Industry Rates, Workplace Rates and the Wage-Drift," 
Manchester School, May 1956. See also B. C. Roberts, "Trade Union Behavior and 
Wage Determination in Great Britain," in Theory of Wage Determination, op. cit. 

22 See E. H. Phelps-Brown and P. E. Hart, "The Share of Wages in National 
Income," Economic Journal, June 1952. 

103 



table 2. Indices of Hourly Money Earnings in Manufacturing 
in Selected Countries, 1946-57 

(1950 - 100) 

19461947194819491950 1951 1952 1953 1954 1955 1956 1957 

France 37 53 81 91 100 128 148 152 162 174 187 202 

Italy — 71 94 99 100 110 115 118 122 129 138 — 

Norway 79 87 92 94 100 114 127 133 140 148 159 169 

Sweden 74 85 93 96 100 121 144 150 156 168 183 — 

United Kingdom 79 87 93 96 100 110 118 125 132 143 155 165 

Germany 70 73 82 94 100 113 122 127 130 139 152 166 

Holland 81 87 92 92 100 108 110 113 132 136 150 — 

United States 74 84 92 95 100 108 114 120 123 130 135 141 

Sources: International Labor Review, Statistical Supplements; and United Nations Sta- 
tistical Yearbooks. 

low (but unmeasured) levels in France. Norway, Sweden, and the United 
Kingdom have all undertaken "responsible" wage policies during part of 
the postwar period, and when responsibility was most in practice, up to 
1950 and the Korean War, wage increases may have been slowed down a 
bit; but their records are not much different from that of the United States, 
where no such policy was in effect. In fact it was in Sweden during this 
early postwar period that the term "wage drift" was invented. 23 

Holland and Germany have had stronger policies of wage restraint in 
the postwar period, for the sake of the restoration of their economies, but 
wages have gone up only somewhat less than in Norway and the United 
Kingdom, and more than in Italy. In Germany, wage restraint was par- 
ticularly in force from 1949 (after currency reform) to 1955 with some 
apparent effect, but this was also a period of great absorption of refugees 
into the economy. In Germany, with wage restraint by the unions, a "wage 
drift" began to show up in pronounced form by 1954 particularly in the 
metal-working industries of North Rhine-Westphaha. And it might be 
noted that a "wage drift" above contract rates becomes increasingly em- 
barrassing to unions and undermines a wage restraint policy. 

Finally, the United States, without wage restraint and with a sectional 
bargaining approach, has demonstrated a comparatively high degree of 
wage stability, as Table 2 shows. 

Russia — Russian statistics on a comparable basis are not readily avail- 
able. However, some comparisons can be made. From 1948 to 1952 
money wages are said to have risen 8 per cent in Russia as against 24 per 



23 For an early use of the term see Rudolph Meidner, 'The Dilemma of Wages 
Policy Under Full Employment" in Wages Policy Under Full Employment, Mac- 
millan, 1952. 

104 



cent in the United States; 24 and from 1953 to 1956 the figures are 7 per cent 
and 12.5 per cent. 25 Also, it should be noted, productivity, as an offsetting 
force, has been rising faster in Russia (though it is at a much lower abso- 
lute level) than in the United States. But even in Russia, with an "agent 
of the state" union movement and authoritarian control, money wages 
have been rising; and price rates have been particularly resistant to controls. 

Observations 

The record, inadequate as it is, does permit some conclusions. 

1. The "class bargainer" (or "enemy of the state") union movement, in 
the type of economy in which it develops, may well add to inflationary 
wage pressures. 

2. The "agent of the state" union movement, in the type of system 
where it finds its natural habitat, is compatible with a comparatively slow 
rate of increase in the general level of money wages. 

3. The "partner in social control" union movement may join in keeping 
wage increases somewhat below their normal levels for relatively short 
periods of time. But the "wage drift" and the internal pressures which 
develop under a wage-restraint policy make it unlikely that this effect will 
be long lasting. The results of wage restraint have been modest at best, 
although useful under the circumstances where they have been applied. 

4. The "sectional bargainer" union movement presents a more mixed 
situation. When in a state of excitement, as around 1937 in both the 
United States and Great Britain, it may push wages up beyond "normal." 
In a depression, it may well hold them somewhat higher than they other- 
wise would be. At the plateau of a period of prosperity or in the early 
downswing, it may continue rates of wage increases experienced in the 
recent and more favorable past into the new situation. But in the upswing 
or a period of demand inflation, it may, as Rees has argued for the basic 
steel industry in the United States from 1945 to 1948, actually retard 
wage increases. 26 

Generally, the "sectional bargainer" union movement will probably lead 
to a steadier advance of the general wage level, neither as fast nor as slow 
as might otherwise occur. Also, through pattern bargaining, wage increases 
may be spread more uniformly and more broadly throughout the economy 
than under non-union conditions. Thus the total long-term effect is likely 



24 J. Chapman, "Real Wages in the Soviet Union, 1928-1952," Review of Economics 
and Statistics, May 1954. 

25 United Nations, World Economic Survey, 1957. 

26 Albert Rees, "Postwar Wage Determination in the Basic Steel Industry," American 
Economic Review, June 1951. 

105 



to be moderately inflationary; for the postwar period in the United States 
a net impact (after allowance for the influence of other factors) on the 
price level of somewhat less than one-half of one per cent as compared 
with "normal" or non-union conditions. 

The real question might be why, as compared with the havoc it might 
wreak as seen by Lindblom, it has had so little effect? The answer must 
lie, in part, in the general reasonableness of the unions and their leaders 
in the context of the type of society in which they evolve; and thus in the 
nature of this kind of union as an institution. 

In fact, two reversals of common statements come closer to illuminating 
the truth. Instead of asking why unions have so much inflationary effect, 
it might be more pertinent to ask why, as "monopolies," they have so little. 
Instead of accusing unions of an effective upward pressure on wage levels 
in a period of expansion and inflation, it would be more pertinent to make 
the accusation about them in periods of depression and deflation. The 
wrong question is asked; and the wrong accusation made. 

5. The volume of unemployment is closely related to changes in the 
general level of money wages. In general, the level of employment must be 
considered the most important single factor. Its influence is over and be- 
yond that of the trade union. 

6. A period of expansion in a capitalist economy is normally a period of 
some inflation. Expansion and inflation are common travelling companions, 
whether a union movement travels with them or not. 

7. Government wage controls can have an effect in holding down wage 
levels, perhaps more in the short run than in the long run, except in an 
authoritarian economy like the Russian. 

8. Administered prices most certainly can make it easier to pass on 
administered wages without affecting profits. 

The type of union and the character of the environment together de- 
termine the impact of the union on the general level of money wages. To 
view either one alone is to view but part of the scene. Taken together, in 
Western capitalism, the combination has probably become a somewhat 
more inflationary one than in earlier times. The union has often become 
more insensitive to the pressure of unemployment because of seniority rules 
protecting its older members and unemployment compensation for its newer 
members; but offsetting this has been the general growth in reasonableness 
and a sense of responsibility. The major changes are in the environment 
which is more permissive — full or more nearly full employment, the spread 
of administered prices and the drying up of pools of readily available labor. 

If the unions secure greater wage increases than in the past, it is not so 
much because they want "more, more and more," which they do, but 
rather because it is easier to get "more, more and more." The environment 
is more conducive, rather than the unions more insatiable. The source of 

106 



the trouble, to the extent there is trouble, is more that there is less pressure 
on the wage fixers than that the wage fixers are less sensitive to it; is more 
that there is less power in the environment and less that there is more 
power in the unions. 

If remedies are to be sought, they would seem to lie, first, in strengthen- 
ing the pressure of the environment toward stability and, second, in making 
unions more sensitive to that pressure. 

Remedies 

In considering remedies, in the context of the American economy, it may 
be well to contemplate these four points: 

( 1 ) Some inflation may be a normal cost of growth; 

(2) The United States has had a comparatively good record on inflation 
in the postwar years; 

(3) Some mild inflationary pressures are inherent in the kind of union- 
ism which evolves out of American society; 

(4) Certain "solutions" are not compatible with the character of this 
society — "agent of the state" unionism, or even "social partner" 
unionism, or permanent unemployment in excess (and possibly sub- 
stantially in excess) of 6 per cent, or, probably, permanent wage 
(and price) controls by government. 

Within the context of our society, however, several things may be 
possible: 

1. To begin with, it would not be wise to guarantee full employment, 
particularly sector by sector, regardless of wage and price behavior. There 
should be some costs to irresponsible actions. 

2. Next, administered prices are not fully socially accepted and their 
more unreasonable excesses should be discouraged by all reasonably avail- 
able means, including anti-trust action and freer trade. 

3. Industries of great pattern-setting importance or otherwise crucial to 
the economy should be made subject to ex post and ad hoc impartial fact- 
finding review of their wage bargains (and price policies) to acquaint the 
public with their consequences. This is one way to mobilize public opinion 
to bring pressure for stability on the private wage and price fixers. 

4. The government should not enter industrial disputes with a "peace-at- 
any-price" approach except in a true national emergency. 

5. All available action should be taken to increase the total supply of 
labor, for example, by providing part-time jobs for housewives and older 
persons, and to improve the mobility and adaptability of the labor force. 

6. Unions should be open to all qualified workers. At the same time, 
rival unionism and great internal union instability should be avoided since 

107 



the conflicts arising from them usually find their solution, in part, in 
wage increases. 

These are reasonable means and only reasonable results should be ex- 
pected from them. We are living in an age marked by uneven but rapid 
economic growth, by a commitment to more-or-less full employment, by 
an exhaustion of earlier available sources of new accessions to the industrial 
labor force, by the great advancement of group initiative and group control 
over the economy, by the substantial freedom of individuals and groups 
from the imposed power of the state, and by mild inflation. Remedies for 
the last phenomenon must be seen in the light of the other phenomena 
which surround it. All things are not possible in all situations; and one 
thing which is not possible in this situation is full price stability and the 
wage levels which are consistent with it. The most successful case of wage 
control in an industralized nation in the postwar period is also the most 
repugnant. 



108 



5 




Wage behavior and inflation: 



An international view 



Lloyd G. Reynolds 



This paper will review the problem of wage-induced inflation in the light 
of experience in Western industrial countries. We shall say nothing about 
experience in low-income, primary producing economies or in the countries 
of the Soviet bloc. It is worth noting, however, that the problem is not 
absent even in centrally planned economies. Throughout the nineteen 
thirties and forties, the rapidly expanding urban industries of Russia 
struggled with a chronic manpower shortage. There was little effective 
restriction of labor mobility, turnover rates were high, and each plant 
manager scrambled for labor as best he could. One way of getting more 
labor was to pay more than the official wage scales. Overpayment was 
common, and the State Bank seems to have been willing to provide extra 
payroll funds when failure to do so would have meant an interruption of 



Author of numerous books and professional articles on economics and administra- 
tion, Lloyd G. Reynolds is Sterling Professor of Economics at Yale University. 
In 1941-43 Dr. Reynolds was regional price director for O.P.A.; in 1943-45, co- 
chairman of the appeals committee of The National War Labor Board; and in 1945- 
47, consultant to the United States Bureau of the Budget. He has also served as 
president of The Industrial Relations Research Association, as vice-president of The 
American Economic Association, and as one of the editorial board of The Ameri- 
can Economic Review. Currently he is a director of The National Bureau of 
Economic Research. 

109 



production. The actual monetary circulation rose considerably faster than 
the planned circulation, increasing the pressure in retail markets and con- 
tributing to the great rise of prices over these two decades. Only in the 
nineteen fifties was the volume of wage payments brought under reasonably 
effective central control. 

The movement of wages and prices over the past decade in certain of the 
Western industrial nations is shown in Table 1. If one excludes France 

table 1. Percentage Increase in Selected Economic Indicators 
in Selected Countries, 1948-1957 





Money 


Wholesale 


Cost Oj 


Real 


Real GNP 




Wages 


Prices 


Living 


Wages 


Per Capita 


Australia 


109 


102 


102 


4 


1 


Canada 


76 


17 


26 


39 


313 


France 


149 


66 


77 


40 


60 


German F. R. 


105 


17i 


14 


78 


752 


Italy 


543 


-1 


28 


203 


70 


Netherlands 


6H 


43 


48 


154 


264 


Norway 


84 


71 


53 


20 


35 


Sweden 


974 


53 


47 


394 


254 


U.K. 


77 


555 


51 


18 


16 


U.S. 


54 


13 


17 


31 


253 



1 1948 estimate based on July-December only. 

2 From 1950 only. 

3 1957 estimated. 

4 Through 1956 only. 

5 Through 1955 only. 

Sources: Money wages, wholesale prices, cost of living, and population from UN Statistical 
Yearbooks, 1955, 1957, and Monthly Bulletin of Statistics, September 1958. 
GNP at market prices from OEEC Bulletin, July 1958, except Australia, which is 
from IMF Financial Statistics, January 1956 and January 1959. 

(with its chronic inflation problem) and Australia (where the economy was 
disrupted by the violent upsurge of world wool prices during the Korean 
War), the consumer price index rose in most countries by from 20 to 50 
per cent. Britain, Norway, Sweden, and Holland all experienced roughly 
a 50 per cent increase, due partly to delayed release of suppressed inflation 
held over from World War II. At the low end, the United States experi- 
enced only a 17 per cent increase and West Germany only 14 per cent. 

Considering the overhang of suppressed inflation from World War II, 
the physical reconstruction effort of the late 'forties, the fresh price upsurge 
set off by the Korean War, and the very high level of investment and output 
in most countries during the 'fifties, this rate of price increase is not sur- 
prising. It should be noted also that most of the increase occurred during 
the first half of the period. During the years of peacetime prosperity 1952- 
57, the cost-of-living index rose by something between 5 and 15 per cent 
in most of the countries studied, or a rate of between 1 and 3 per cent per 

110 



annum. This rate of increase may well be considered undesirable, but it is 
by no means abnormal. Table 2 shows the annual rate of price increase 
associated with periods of economic expansion in selected countries since 
1870 or so. It is clear that output expansion has typically been associated 
with a rising price level, and that there is nothing novel about the 1950's 
in this respect. 

table 2. Industrial Production and Prices During Periods of 
Rising Output, by Country 

{Average annual percentage change) 







Wholesale 


Consumer 


Country and period 


Production 


Prices 


Prices 


Germany 








1880-1891 


6.8 


0.2 


• • • 


1892-1899 


9.0 


0.6 


• • • 


1901-1907 


6.1 


3.2 


2.4 


1907-1913 


6.4 


1.5 


2.0 


1924-1929 


9.8 


• • • 


4.1 


1932-1938 


20.4 


1.6 


0.8 


1952-1957 


11.6 


0.4 


0.9 


France 








1879-1883 


6.7 


-1.5 


• • • 


1886-1890 


6.0 


1.3 


a • • 


1895-1900 


4.6 


3.3 


• • • 


1902-1907 


4.3 


3.2 


• • • 


1908-1913 


6.2 


3.0 


3.0 


1923-1929 


8.5 


7.6 


13.4 


1932-1937 


4.0 


7.6 


4.4 


1952-1957 


8.5 


0.6 


0.8 


Sweden 








1870-1876 


12.3 


1.1 


• • • 


1879-1885 


9.1 


-1.5 


• • • 


1888-1898 


18.6 


-0.4 


• • • 


1901-1907 


5.0 


2.0 


1.7 


1909-1913 


7.2 


3.1 


1.3 


1923-1929 


9.8 


-2.2 


-0.7 


1932-1937 


14.1 


5.1 


0.6 


1952-1957 


3.8 


0.4 


3.1 


United Kingdom 








1870-1874 


3.9 


1.6 


1.1 


1879-1883 


7.9 


... 


0.4 


1886-1890 


5.2 


0.9 


• 


1893-1899 


4.1 


• • • 


-0.6 


1904-1907 


4.0 


4.4 


1.1 


1908-1913 


6.2 


3.3 


1.9 


1923-1929 


2.7 


-2.4 


-0.8 


1932-1937 


8.5 


5.3 


1.4 


1952-1957 


4.3 


. . • 


3.7 



111 



Table 2 — Continued 







Wholesale 


Consumer 


Country and period 


Production 


Prices 


Prices 


United States 








1876-1882 


12.0 


-0.3 


. . . 


1885-1892 


10.0 


-1.5 


... 


1896-1903 


11.0 


4.0 


2.2 


1904-1907 


9.9 


3.1 


3.2 


1908-1913 


9.8 


2.2 


2.4 


1924-1929 


7.0 


-0.6 




1932-1937 


20.0 


6.6 


1.0 


1952-1957 


5.0 


2.2 


1.5 



Source: United Nations, World Economic Survey, 1957, pp. 19-20. 

Money wages rose faster than prices in every country in 1948-57, 
as would be expected in a progressive economy where rising productivity 
serves as a partial offset to wage gains. Real wages thus rose substantially 
in every country except Australia. Excluding Australia on the low side 
and West Germany on the high side, real wages in most countries rose by 
between 20 and 40 per cent, or roughly 2 to 4 per cent per year. The 
greater part of this increase came in the latter half of the period, after 
postwar reconstruction had been substantially completed and the disloca- 
tion of the Korean War had subsided. 

There is little relation between the rate of increase in money wages and 
real wages. Some countries with relatively small money-wage increases 
show large increases in real wages, and vice versa. There is a much closer 
relation between the last two columns of Table 1 — increase in real-wage 
level and increase in real per capita output of the economy. (There is of 
course no reason why the two figures should move precisely together. The 
GNP figure is calculated per head of population, and includes government 
outlay and capital formation as well as personal consumption. The real 
wage figure is per hour of time worked, and for most countries covers only 
manufacturing industry. One would expect a rough relation between the 
two, however, and this is what we actually find.) Real wages rose most 
rapidly in West Germany because of the phenomenal increase of production 
in that country. They rose only moderately in Britain, where the produc- 
tion increase over the decade was also quite moderate, and in Australia per 
capita output and real wages increased scarcely at all. The productivity 
basis of real-wage gains stands out quite clearly from the data. 

A country-by-country review of experience over the past decade would 
be tedious and difficult to digest. It seems better to work through the 
problem of cost inflation in an analytical way, bringing in data from 
individual countries where they serve a particular purpose. We shall do 
this under the following headings: 

( 1 ) The mechanism of cost inflation. 

112 



(2) Statistical indicators of cost inflation. 

(3) Structural determinants of cost inflation: the labor market, indus- 
trial structure, worker attitudes, union behavior, and the collective 
bargaining system. 

(4) What to do about it? 

The mechanism of cost inflation 

In Europe, as in the United States, prices and wages have been rising 
hand over hand for the past twenty years. The statistical measurements, 
however, provide no direct clue to the underlying sources of inflation. 
Is it an excess of money demand which has been pulling up prices and 
wages? Is it unreasonable wage increases which have forced up both 
prices and the monetary circulation? Or what? This familiar conundrum 
must be faced at the outset. 

Three types of inflation 

A rise in the price level may come about in at least three ways: 

(1) An increase in total money demand to the point where it exceeds 
the productive capacity of the economy at current prices. Consumers, 
investors, and government together are trying to buy more goods than the 
economy can provide. Prices must rise and real demand must be cut back 
(or, alternatively, a system of physical rationing must be imposed) until 
demand once more fits within the limits of capacity. This is the traditional 
case of demand inflation, which appears most clearly during periods of 
war and defense mobilization. 

(2) An important variant of this case occurs when aggregate demand, 
while not excessive in terms of total output capacity, exceeds the 
capacity of one or more key industries which thus become bottlenecks in 
the expansion. The structure of money demand does not correspond with 
the structure of capacity. Thus while there is still under-utilization in many 
industries, the bottleneck industries find themselves booked to capacity or 
beyond. Their prices and profits rise, wage increases are readily granted; 
and this tends to induce price and wage increases elsewhere in the economy. 

It is this situation, rather than a general excess of demand, which com- 
monly characterizes the latter phase of a peacetime boom. Consider, for 
example, the experience of Western Europe and North America during the 
middle 'fifties. The inflationary forces released by the Korean War were 
largely spent by 1952. The years 1953 through 1957 were in most coun- 
tries years of high peacetime prosperity. Except perhaps in France and 
the Netherlands, there was little indication of anything which could be 
called aggregate excess demand. There was, however, heavy pressure on a 

113 



number of raw materials and durable goods industries, leading generally 
to shortages, order backlogs, increased costs and rising prices. The upward 
tendency of coal, steel, and machinery prices in selected countries is shown 
in Table 3. These prices typically rose considerably more than the general 
wholesale price level, indicating that excess demand was concentrated in 
the durable goods sector. 

In demand inflations of types (1) and (2), costs are clearly not the 
aggressor, but this does not mean that their behavior is unimportant. On 
the contrary, the responsiveness or flexibility of costs in the face of price 
increases will have an important influence on the duration and height of 
the price movement. A long lag of wages behind prices will have a 

table 3. Wholesale Prices of Coal, Steel, Machinery, and All 
Commodities, Selected Countries, 1957 





(1953 = 


100) 






All 


Country 


Coal 




Steel 


Machinery 


Commodities 


Canada 


109 




— 


— 


103 


Federal Republic of Germany 


112 




105 


106 


105 


France 


103 




114 


— 


108 


United Kingdom 


140 




122 


115 


— 


United States 


116 




119 


121 


107 



Source: First three columns from United Nations, World Economic Survey^ 1957, p. 33. 
All commodities index from sources cited in Table 1. 

dampening or "braking" effect. Conversely, automatic escalator clauses 
or other measures to speed up wage adjustments will add momentum to the 
inflation. A long wage lag which cuts workers' purchasing power may well 
be regarded as inequitable and unfeasible, but there is little doubt of its 
stabilizing character. 

(3) "Cost inflation" exists when there is no excess of aggregate demand 
or sectoral demand, and when pressure on prices is being exerted solely 
from the cost side. This can happen in a variety of ways. In a country 
which depends heavily on foreign trade, a rise in import or export prices 
can set off a cycle of increases in living costs and wage rates. This is a 
serious problem in underdeveloped countries, but is found also in more 
advanced economies such as Norway, where imports form close to half of 
net national product. In Norway a period of near stability in prices during 
the late 'forties was violently disrupted by sharp increases in raw materials 
and other import prices after the outbreak of the Korean War. This 
pushed up the level of finished-goods prices, which stimulated substantial 
wage demands, which further increased the prices of domestic products. 
This spiral continued, because of the time lags involved, until well after 

114 



the world boom in raw-material prices had subsided. By 1953 the cost-of- 
living index had advanced 36 per cent, and average hourly earnings in 
manufacturing 44 per cent over the 1949 level. 1 The raw-materials price 
movements set off by the Korean War led to similar spirals in other West- 
ern European countries and, on a smaller scale, in the United States. 

Another type of cost pressure is exemplified by the abandonment of 
wartime price restraints and consumption subsidies in many countries 
during the late 'forties and early 'fifties. This raises the cost-of-living 
index, which stimulates wage demands, which lead to further price in- 
creases, and so on. Retail food prices in the United Kingdom, for example, 
rose about 50 per cent between 1950 and 1956 although prices paid to 
farmers in Britain rose only about 20 per cent. The main reason was that 
food subsidies were reduced and state trading was abandoned, so that food 
imports entered Britain at world prices rather than at specially negotiated 
prices. This had a substantial inflationary effect. In most countries rents 
also rose much more than other living-cost items during the early 'fifties as 
controls were reduced and rents returned to free market levels. 

Finally, cost inflation may originate in an effort by some important bloc 
in the economy to improve its position relative to other groups. If farmers, 
trade unionists, manufacturers, or retailers attempt to better their position 
by manipulating prices or wages, the price level may be pushed upward 
with no ready check. If the economy is operating at full employment, 
moreover, such a movement may easily become cumulative. Increased 
money demands by one group will soon be matched by increased demands 
by others. All may end up in about the same relative position as regards 
real income, but at a higher price level. Frustrated in the attempt to 
appropriate more real income, one or more groups may increase their 
money demands again and set off a new turn of the spiral. To speak of 
"turns" or "rounds" is of course to speak metaphorically, for when such a 
process is in motion the interaction of prices is complex and continuous, 
and one can no longer distinguish cause and effect in a temporal sense. 

The feasibility of cost inflation 

Economics is a tradition-bound subject, and economists have been 
slower than practical men to recognize the possibility that the price level 
may be forced up from the cost side. This springs partly from the habit 
of theorizing about a competitive economy in which price and wage 



1 This was of course not the only factor involved in the Norwegian situation, but 
was the dominant factor during this period. See Mark W. Leiserson, Wages and 
Economic Control in Norway, 1945-57 (Cambridge: Harvard University Press, 
1959), Chap. 5 and 6. 

115 



manipulation is impossible, partly from traditional models of inflation in 
which the impetus comes always from increased spending. 

It has even been denied that there can be such a thing as "cost-push 
inflation" independent of monetary developments. 2 The reasoning proceeds 
as follows: Suppose that unions enforce wage demands greater than those 
warranted by productivity increases, and suppose that processors and 
distributors cover these cost increases by appropriate price mark-ups. 
Where will the money come from to support the higher price level? If the 
monetary authorities hold the monetary circulation unchanged, demand 
will be inadequate to buy the previous output at the new price level. Pro- 
duction and employment will fall, and this will either compel price and cost 
reductions or at any rate act as a brake on further increases. Suppose on 
the other hand that the monetary authorities, fearful of bringing on a 
recession, permit whatever increase in money supply may be necessary to 
support the higher price level. Then the new price level is feasible, and 
can be raised step by step for as long as the permissive monetary policy 
continues. But does this not mean merely that the monetary authorities 
have been bullied into creating an excess of money demand? If so, do we 
not end up with the conclusion that all inflation is after all demand in- 
flation? It would then follow that inflation remains a monetary problem 
to be dealt with by the usual instruments of monetary policy. 

There are two main difficulties in this argument. First, it is not true 
that there is a close relation in the short run between aggregate money 
supply and aggregate spending. Changes in monetary velocity or liquidity 
have been revealed as of great practical importance during the postwar 
period. In some European countries, to be sure, the postwar inflation has 
been financed mainly by expansion of the money supply. France and Italy 
are the outstanding examples. But in others, including Denmark, Sweden, 
the Netherlands, and the United Kingdom, price increases have been 
financed mainly by a reduction of liquidity, with the money supply rising 
little or even contracting. 

In the Netherlands, for example, the increase in money supply averaged 
only 1.6 per cent per year over the period 1948-56. Considering that the 
increase in physical output averaged 5.4 per cent per year, this was a 
very modest rate of monetary expansion and would suggest a decline in 
the price level. Actually, the price level rose at an average rate of 3.8 per 
cent per year. How could this happen? The explanation is that monetary 
liquidity declined consistently and in some years quite sharply through- 
out the period, except during 1952 and 1953. Over the nine years as a 
whole, increased velocity added the equivalent of 5.7 per cent per year to 



2 See for example Walter A. Morton, "Trade Unionism, Full Employment, and In- 
flation," American Economic Review, March 1950, pp. 13-39. 

116 



money supply, or more than three times the rate of increase in actual 
monetary units. 3 

This element of "play" in the system means that the monetary brakes 
on cost inflation are not nearly so tight as they otherwise would be. It can 
still be argued that, if velocity is increasing more rapidly than monetary 
requirements based on a stable price level, the monetary authorities can 
always offset this by producing an appropriate decline in money supply. 
This increases the trickiness of monetary policy, however, particularly since 
the rate of increase in velocity is variable and difficult to predict. 

Waiving this difficulty, however, there is a more fundamental question 
about the monetary line of argument. What good does it do to re-label 
price increases arising from cost pressure as "demand inflation"? How does 
this change the actual situation? It is formally correct that a certain price 
level is feasible only if adequate monetary demand is available, and in this 
sense any inflation must be a demand inflation. But this does not change 
the nature of the phenomenon. It does not alter the fact that producer 
groups can and may demand monetary returns incompatible with stable 
prices, and that this creates a special kind of problem for the monetary 
authorities, differing materially from the traditional problem of restraining 
excessive aggregate demand. Indeed, it may turn out that monetary policy 
is incapable by itself of coping with cost pressures. We shall return to this 
issue at a later stage. 

How much is "too much"? 

While cost pressure may emanate from any producer group, we are 
mainly concerned in this paper with pressure arising from excessive wage 
increases. Here we face the initial difficulty that a rising money- wage level 
based on productivity increases is normal in a progressive economy. 
Money-wage increases are disruptive only if they rise above some "toler- 
able," "warranted" or "non-inflationary" rate. There is an initial problem, 
then, of defining what one means by a non-inflationary behavior of wages. 

There are at least two possible approaches to this problem, which we 
may term the productivity approach and the income approach. The former, 
which typically involves comparison of the rate of increase in hourly wage 
payments and in output per man-hour, has serious limitations for the 
present purpose, and we must examine briefly why this is so. The year-to- 
year increase of output in a growing economy arises from three sources: 

(a) An increase in man-hours of labor utilized in production. This 
factor is eliminated when total national output is divided by man-hours 



3 Robert Triffin, "Credit, Money, Production, Prices and Balance of Payments in 
O.E.E.C. Countries, 1948-56," (Paris: unpublished OEEC memorandum, February 
1958), p. 49. 

117 



worked, and the increase in output per man-hour is attributable to the two 
remaining factors. 

(b) An increase in physical capital used in production. The fact that 
in the United States the supply of capital has risen considerably faster 
than the supply of labor since 1900, so that capital per worker has in- 
creased about 1 per cent per year, is partly responsible for the rapid rise 
of man-hour output. 

(c) A residual not explained either by the increase of labor or of 
capital and which, since we cannot explain it, we sweep under the rug by 
ascribing it to "technical progress." About half the increase in the national 
output of the United States since 1900 falls in this residual category. 4 

The proposition that wage rates or earnings per man-hour should rise 
at the same rate as output per man-hour is in no way self-evident. On the 
contrary, it involves complex assumptions about the operation of a market 
economy with changing factor supplies, and about equity in income dis- 
tribution, which cannot be detailed here. Moreover, this or any similar 
proposition is a statement about the normal or desirable distribution of 
real income between capital and labor — a classic issue in economics, but 
one rather remote from the problem of inflation. A practical difficulty is 
that our measures of man-hour output are very incomplete. The figure 
most commonly cited — man-hour output in manufacturing — covers only 
about one-quarter of the economy, and has almost certainly risen more 
rapidly over the last century than man-hour output in general. 

For these reasons the income approach seems to have a more direct and 
useful bearing on the matter of inflation. Rising national output over the 
course of time creates what the Swedish economist Erik Lundberg has 
termed "wage space" — room for an increase in money incomes without 
any upward adjustment of the price level. As a first approximation, one 
can define a non-inflationary situation as one in which wage income per 
capita is rising at the same rate as real national output per capita. (A more 
refined statement would have to specify the behavior of other types of 
money income, and also the behavior of savings and investment.) For the 
United States, Kuznets has estimated this at about 16.4 per cent per decade 
over the period 1894-1954, but the rate of increase since 1940 has been 
considerably higher than in earlier years. 5 

This figure in some ways overstates, in other ways understates, the 
feasible rate of increase in basic wage schedules. On the side of over- 
statement we may note: 



4 See Solomon Fabricant, Basic Facts on Productivity Change (New York: National 
Bureau of Economic Research, 1958), Occasional Paper 63. 

5 Simon Kuznets, "Quantitative Aspects of the Economic Growth of Nations," 
Economic Development and Cultural Change, V, No. 1 (October 1956), p. 10. 

118 



1. Part of the increase in national output comes about through the 
transfer of workers from low-productivity to high-productivity industries, 
notably from agriculture to manufacturing, over the course of economic 
development. This increase in output is presumably offset by increased 
earnings of the transferred workers in their new occupations. To use it 
also as a basis for a general increase in wage schedules would involve 
double counting. Increases in output from this source should be deducted, 
therefore, in calculating the feasible rate of wage increase. 

2. It is the rate of increase in earnings which should correspond to the 
rate of increase in output. Earnings, however, may rise considerably faster 
than basic wage rates, particularly during a period of sustained high 
employment. The reasons include promotion and upgrading of individual 
workers, overrating of jobs to permit higher wage offers for recruiting 
purposes, loosening of time standards on piece work so that earnings pull 
farther and farther above base rates, and straight overpayment of the union 
scale. This tendency for earnings to diverge farther and farther above 
official wage schedules, which Swedish economists have termed "wage 
drift," has been noticeable in most industrial countries during the 'forties 
and 'fifties. 

3. Workers' incomes have also tended recently to rise faster than basic 
wage rates because of the rapid growth of fringe benefits. These payments 
form a rising proportion of workers' total compensation, and direct wage 
payments a declining proportion. 

Two important considerations may be advanced in the opposite direction: 

1. Why should all types of income in the economy move upward in 
lock step? Perhaps wage earners' incomes can rise faster than per capita 
output if other people's incomes are rising less rapidly. Perhaps workers 
may be able to get "more than their share" of a rising national income at 
the expense of other groups. There is doubtless something to this line of 
argument. If one defines "labor" narrowly to include only wage earners, 
then wages may gain somewhat at the expense of salaries over the long run. 
Another "squeezable" group includes recipients of rents, fixed interest 
payments, annuities, and pensions. But these groups receive only a minor 
part of national income, and their squeezability is limited and tends to get 
used up over the course of time. One cannot, therefore, rely on income 
redistribution to allow workers' incomes to rise much faster than per capita 
national output is rising. 

2. Another consideration of some importance is the behavior of labor- 
force participation rates and weekly or annual hours of work. If these two 
variables are increasing, then hourly wage schedules should rise less rapidly 
than the desired rate of increase in wage income per capita, and vice versa. 
In the United States, the overall labor-force participation rate has been 
remarkably stable since 1900, while the trend of hours has been steadily 

119 



downward. This would indicate that hourly wage rates should rise faster 
than per capita wage income. 

Taking all these things into account, it is possible in principle to define 
a non-inflationary rate of increase in basic wage schedules. Quite elaborate 
calculations would be necessary, however, to get even a rough idea of 
what this rate may be in a particular country at a particular time. It is 
commonly asserted that in the United States at present the noninflationary 
rate of increase in wage schedules lies somewhere between 2 and 3 per cent 
per year. This is probably correct, but has not yet been confirmed by 
rigorous investigation. 

The concept of a feasible, or reasonable, or non-inflationary rate of 
increase in the money wage level is obviously useful. But it is also a 
dangerous concept, because it is apt to be interpreted as meaning that 
every wage rate in the economy should rise by a certain percentage every 
year. This never happens, and it would be unfortunate if it did happen. 
Wages should rise faster than average in expanding companies, industries, 
and regions, and less than average in stationary or declining areas and 
industries, in order to assist reallocation of the labor force to the growth 
points of the economy. Apart from this, the wage structure always contains 
inequities and anomalies which can be corrected only if there is flexibility 
for different wages to advance at different speeds. It may also be undesir- 
able for the average wage level to advance at the same rate from year to 
year. Subnormal wage increases are natural in recession years, while in 
boom years the rate of increase may be unusually high. The feasible rate 
of wage increase over the long run should not be construed as a yardstick 
which can be applied inflexibly to particular wage decisions. 

A possible mechanism of wage inflation 

Does the actual rate of money-wage increase tend to rise above the rate 
compatible with price stability, particularly at high levels of output and 
employment? Factual evidence is difficult to come by, but one can think 
of reasons why this might happen. 

One plausible line of reasoning takes off from the fact that, in a dynamic 
economy, industries differ widely in their rate of productivity increase. 
Calculations for eleven countries over the period 1950-56 show that the 
standard deviation of individual industry rates of increase in man-hour 
output was typically between 80 and 100 per cent of the mean rate of 
increase for manufacturing as a whole. 6 The main reason seems to be 
inter-industry differences in the rate of growth of total output. Young, 
expanding industries with a rapid rate of growth in demand and production 



6 United Nations, World Economic Report, 1957, pp. 36-37. 

120 



are able to achieve much larger gains in man-hour output than mature 
industries with little output expansion. Thus in Canada all manufacturing 
industries showed an annual increase of 2.4 per cent in man-hour output 
over the years 1950-56; but the top quarter of industries in terms of out- 
put growth showed an average annual increase of 6.8 per cent in man-hour 
output. The corresponding figures for Sweden are 2.8 per cent for all 
manufacturing, 6.4 per cent for the quarter of most rapidly expanding in- 
dustries; for the Netherlands, 2.6 per cent and 4.8 per cent. Less striking 
differentials in the same direction appear in all the other countries studied. 

A related type of evidence has to do with whether rising wage rates are 
accompanied by rising or falling wage costs. Over the period 1950-56 
covered by the U. N. study, money wages were rising at a substantial rate. 
Yet in most countries unit wage costs fell in the metals, machinery, petrole- 
um, and coal products industries. Unit wage costs held about even in 
chemicals, rubber, glass, paper, and wood products industries. Unit wage 
costs rose in most countries in textiles and clothing, leather and leather 
products, tobacco products, and food processing. In the mature consumers' 
goods industries in this last group, the rate of output growth and pro- 
ductivity increase was not sufficient to offset rising wage rates. 

Wage rates in all industries advance at a uniform rate — specifically, at 
the non-inflationary rate defined above. Industries with a low rate of 
productivity increase will then find their unit wage costs rising, and these 
cost increases will have to be covered by price increases. At the extreme, 
in service industries such as barbering where most cost is labor cost and 
there is little room for productivity gains, product prices will rise pari passu 
with the general wage level. Industries with high rates of productivity 
increase, on the other hand, will find their unit labor costs falling and can 
afford price reductions. Given the pervasive influence of oligopoly and 
the business antipathy to price cutting, however, it seems unlikely that 
reductions will actually be made on a scale sufficient to offset the price 
increases in other industries. 7 Thus even under these conditions of "ideal" 
wage behavior, there is a certain inflationary bias which might cause a 
gradual increase in price levels. 

In actuality, however, it is unlikely that wages will behave in this way. 
Wages will probably advance, or try to advance, at a higher than average 
rate in the industries where man-hour output is advancing most rapidly. 
We have already noted that these tend to be the industries in which total 
output and employment are also rising most rapidly. It may thus be neces- 



7 Where, then, will the savings from productivity increases go? Perhaps partly into 
profit, most of which will be reinvested in further expansion. But partly also into 
improved product quality — a price cut of sorts, but one which does not show up in 
cost-of-living indexes; and partly into a generous policy on wage rates and other 
factor prices. 

121 



sary to raise wages for recruitment purposes, particularly if the economy 
is near full employment. Further, with output expanding and prices diverg- 
ing above the declining trend of unit costs, these industries will be in a 
position to afford a generous wage policy. There is considerable evidence 
that profitable companies and industries tend to share their prosperity with 
employees, even in the absence of union pressure. 8 If a union is present, 
it will reinforce management's good intentions and may press for even 
larger increases than would occur otherwise. 

The pace of wage advance in the lead sectors of the economy will then 
be transmitted to other sectors through both institutional and market 
channels. To the extent that other industries are unionized, and to the 
extent that there is inter-union rivalry on the wage front, there will be 
pressure on employers generally to match the "pattern" established in the 
high-productivity industries. But one does not have to rely on collective 
bargaining to attain this result. Under high-employment conditions, each 
employer must keep reasonably well in step with the general pace of wage 
advance in order to replace or expand his labor force. The market sets 
limits, albeit rather rough ones, to the feasible size of wage differentials; and 
as the top of the wage structure rises, the bottom tends to be pulled up 
after it. 

While it seems plausible that things may work in this way, it is hard 
to check the argument by reference to statistics. If the transmission of 
wage impulses is rapid and complete, one may find that after a certain 
period of time all industries have gone up by roughly the same amount. 
(Indeed, the U. N. report cited previously did find this for the period 
1950-56. In most countries the percentage increase in wages for the top 
quarter of most rapidly growing manufacturing industries was almost 
identical with that for all manufacturing.) This in no way disproves the 
hypothesis that the expanding and profitable industries may have been 
forcing the pace of wage advance. But neither does it demonstrate that 
this has happened, or give any quantitative indication of the degree of 
acceleration. Detailed analysis of the timing of wage adjustments, industry 
by industry, might be more revealing, but little evidence of this sort has 
been collected. 

Statistical indicators of cost inflation 

To this point we have tried to distinguish cost inflation from other types 
of inflationary process, to define the rate of money-wage increase which is 
feasible without producing upward pressure on prices, and to describe a 



8 This was suggested some years ago by Sumner Slichter, and has been confirmed by 
subsequent studies. See S. H. Slichter, "Notes on the Structure of Wages," Review 
of Economics and Statistics, Vol. 32, 1950, pp. 80-91. 

122 



mechanism which might cause actual wage increases to exceed the feasible 
rate — a mechanism which is not dependent on, though it might be 
strengthened by, the presence of unionism and collective bargaining. But 
all this is at a theoretical level. What we really want to know is whether 
cost inflation is a serious problem in actuality. Given an economy in 
which wages, prices, profits, and other monetary quantities are rising 
simultaneously, how does one tell where the inflationary impetus is coming 
from? Specifically, is there any way of telling whether a "wage push" is 
or is not present? 

Some of the measurements commonly used are not very helpful in this 
respect. It proves little, for example, to compare the movement of wage 
incomes and physical output. The conceptual difficulties already noted 
limit the validity of any simple comparison. More fundamentally, however, 
a more rapid increase of wage incomes than of physical output would be 
found in both demand-pull and cost-push situations. It thus cannot provide 
a criterion for distinguishing between them. 

It is not much use either to work with leads and lags, to try to determine 
whether wages rose "before" prices or after them. This would work only 
if the economy had been for some years in a stable state, with little move- 
ment of price indexes and with employment and output expanding steadily 
along a long-term growth path. If after such a period of stability one 
found an upsurge of certain prices followed by abnormal wage increases, 
or vice versa, one might try to analyze causation in a temporal sense. 
But such a stable economy is an economist's construct, not something 
which one can hope to find in actuality. What one does find is continual 
fluctuation of individual prices and wages, and wave-like movements of 
output and employment, the whole never coming to anything like a condi- 
tion of rest. Whether one event occurred "before" another, whether wages 
have risen "more than" prices or vice versa, thus turns entirely on the 
choice of a base period, on where one chooses to cut into the historical 
sequence of events. It is always possible to show that one economic 
quantity has moved "disproportionately" to another by choosing the right 
base, but such arguments obviously have no scientific usefulness. 

There are, however, at least three types of quantitative analysis which 
may throw significant light on the nature of an inflationary process: 

(1) The behavior of profit margins per unit of output or sales. If 
product prices are being pulled up in advance of unit wage costs, then 
profit margins should be widening, and this is in fact a common character- 
istic of demand inflations. If, on the other hand, an increase in unit wage 
costs is the active force, one might expect profit margins to be stationary 
or even f ailing. 

This apparently simple criterion is not without difficulties. First, it is a 
micro-economic criterion which needs to be applied at the industry or 

123 



even the company level. Changes in factor shares of income produced for 
the economy as a whole are not very useful because of their aggregative 
character, because they are usually annual and thus rather blunt for the 
present purpose, and because of well-known difficulties of concepts and 
measurements. Second, while a decline in profit margins can be taken as 
indicating some type of cost inflation, this need not necessarily be wage- 
cost inflation. Speculative increases in raw-material prices, for example, 
such as occurred in the early stages of the Korean War, may for a time get 
ahead of increases in finished-goods prices, and profit margins in some 
sectors may shrink on this account. 

(2) The output level of the economy relative to full capacity, and the 
direction in which output is moving. If output is pressing against capacity, 
and rising gradually as the expansion of capacity permits, an observed 
increase in price levels may reasonably be ascribed to excess demand. If, 
on the other hand, one sees prices rising at a time when output is either 
falling or well below the capacity of the economy, one may suspect some 
degree of cost inflation. 

Here again we need more refined measurements than have been used to 
date. The level of full-time unemployment, which is often used as an 
indication of the gap between output and capacity, is a very crude measure 
indeed. The size of the labor force is somewhat responsive, and hours of 
work are very responsive, to fluctuations in the demand for labor. An 
observed fluctuation of full-time unemployment between 2 per cent and 
8 per cent of the labor force may mean a much larger fluctuation in the 
gap between output and capacity. Moreover, the economy's capacity is 
bounded over any short period by the stock of capital equipment as well 
as by labor supply. One may reach full use of equipment, at least in certain 
bottleneck industries, before labor supply is exhausted. 

It must be remembered, too, that there is considerable lag in the move- 
ment of finished-goods prices, particularly at the retail level. Retail price 
indexes typically lag cyclical downturns and upturns in output by six 
months or so. The fact that prices continue upward for a time after output 
has turned downward is not necessarily evidence of cost inflation. It be- 
comes so only if the divergent movement of prices and output continues 
beyond some "normal" or "reasonable" period of time. 

(3) The detailed character of wage movements. During the postwar 
years Sweden, Norway, and a number of other European countries have 
observed a phenomenon generally termed "wage drift" — a more rapid in- 
crease in actual earnings than in the basic wage rates specified by union- 
management agreements. This comes about through voluntary overpay- 
ment of union scales by employers, individual upgrading, merit increases, 
increased use of incentive systems, loosening of piecework standards, in- 
creased overtime, and other decisions at the plant level. Behind it lies a 

124 



high demand for labor, stemming from excess demand in product markets, 
which both enables and forces employers to follow a liberal wage policy. 

In a year of large "wage drift," one might conclude that unions have 
underpriced their labor and that the effective wage level is being pulled 
upward from the demand side. If on the other hand one finds large basic 
wage increases accompanied by little or no wage drifting, one might sur- 
mise that excess demand is not present. In a very decentralized bargaining 
system such as that of the United States, with many industries only partially 
unionized, one might even find at times a "negative wage drift" involving 
(a) raggedness in the movement of contract rates in different sectors, with 
the high-productivity-increase sectors moving up faster than others; (b) 
raggedness within individual industries, with nonunion companies tending 
to lag behind union companies; (c) a tendency for earnings to advance less 
rapidly than basic rates. This would suggest a cost-push situation in the 
sense that some contractual rates are being pushed up more rapidly than 
is warranted by demand levels in the economy as a whole. 

One could perhaps make a case that price and wage movements, industry 
by industry, should be more nearly uniform under conditions of excessive 
aggregate demand than under cost-push conditions. If so, something could 
be learned from the dispersion of wage and price movements during a 
particular period. Once more, however, one encounters the perhaps in- 
superable difficulty that the results will depend entirely on the time 
period chosen. 

What this comes down to is that aggregative data are virtually useless 
in trying to separate demand and cost inflation. The useful lines of work 
involve micro-economic data, laborious tabulations and analyses, and 
sophisticated judgment in interpreting the results. To do this even for one 
country for a short period is a formidable task. At an early stage of this 
paper I thought that it might be possible to make such an analysis for at 
least a few countries, and to say that cost inflation was or was not important 
during certain periods; but it quickly became apparent that this was a 
subject for a treatise, not an essay. 

Structural determinants of cost inflation 

The fact that wage-push inflation is a hypothetical possibility need not 
mean that it will actually occur. It may occur in some economies at some 
times, but not in other times and places. What determines the susceptibility 
of an economy to wage inflation? The following structural characteristics 
appear to be particularly important: the flexibility of the labor force and 
the efficiency of labor markets; competitive relations and pricing practices 
in product markets; worker attitudes and expectations in the matter of 
wages; and certain characteristics of union organization and collective bar- 
gaining arrangements. 

125 



Labor mobility and labor markets 

The important considerations here are the willingness and ability of 
workers to shift freely among employers, industries, and geographical areas 
in response to economic inducements; and the existence of clearing-house 
arrangements to facilitate the transfer and placement of labor. High pro- 
pensity to move and effective clearing-house arrangements make for a 
flexible labor force, for moderately sloped rather than steeply sloped 
labor-supply curves. 

This is important in two respects. First, the greater the flexibility of the 
labor force, the higher is the level of employment attainable before the 
economy encounters a general shortage of labor and enters an inflationary 
phase. In an economy with a very flexible labor force it may be feasible to 
define "full employment" as corresponding to 2 per cent of the labor force 
unemployed. In an economy with a sluggish labor force and poor labor 
markets, it may not be feasible to reduce unemployment below 5 per cent 
without inducing price inflation. 

Second, we noted earlier that a demand inflation may occur because an 
economy, while still operating below capacity in most sectors, encounters 
production bottlenecks in a few key industries. We ordinarily think of 
these bottlenecks as arising from capacity utilization of plant and equip- 
ment. But the impediment to production may also be a shortage of labor in 
particular industries and localities, even though labor supply is generally 
adequate. This may induce substantial wage increases at the bottleneck 
points, which are then transmitted through market and institutional chan- 
nels to other types of employment. The greater the flexibility of the labor 
force, the less likely is it that such labor bottlenecks will occur; and where 
they do occur, it will not require such large wage increases to overcome 
them as would be necessary otherwise. 

In both respects, then, a flexible labor force raises the production ceiling 
of the economy. It permits rising demand to carry production and em- 
ployment to a higher level before demand becomes excessive and inflation 
sets in. 

Competitive and pricing arrangements 

The customary argument here is that monopoly, "cooperative" oligop- 
oly, and cartel arrangements are favorable to cost inflation, while com- 
petitive pricing acts as a restraining force. This hypothesis may well be 
correct, though not for the reason most commonly offered. There is little 
indication that monopolistic sellers are directly responsible for inflation by 
seeking to expand profit margins more rapidly than other factor costs. On 
the contrary, profit mark-ups seem to be governed by conventional rules 
which remain stable for long periods of time. The UN analysis of inflation 

126 



in the Western European economies over the period 1952-57 found that 
profit margins had not risen more on the average in "administered price" 
industries than in industries where competitive pricing prevailed. 

For all countries as a group, gross margins between prices and direct costs 
did not show any greater tendency to widen in the industries that are commonly 
considered to have administered prices than they did in other industries; in fact, 
during the latter part of the period (1950-56) there was, if anything, a balance 
in the other direction. In only one of the 'administered price' industries did 
margins widen in more countries than they narrowed; this was petroleum and 
coal products, and in the particular period under review the widening margins 
may have been due to changing technology and product mix. Nor does it 
appear that there was any widespread tendency for margins to widen in the 
industries where prices increased most. Industries where widening margins were 
associated with increasing prices or narrowing margins with decreasing prices 
were no more common than industries showing the reverse relationship. 9 

It may be, however, that employers in an administered-price industry 
concede wage increases more readily than they would under free market 
pricing. The administrative ease of converting higher costs into higher 
prices, and the certainty that this can be done fully and rapidly, may lower 
employer resistance to wage increases, whereas uncertainty about the 
product market would have led employers to put up a harder fight. The 
argument is not that administered-price industries raise wages faster than 
other (competitive) industries in the same economy over a given period. 
This is by and large not true, and cannot be true because of the labor 
market linkages explained earlier. The argument is rather that administered- 
price industries raise wages more rapidly than those same industries would 
have done if organized on a competitive basis. If these industries happen 
also to be those in which productivity is advancing most rapidly, they may 
generate a rate of wage increase which is inappropriate for the economy 
as a whole and yet which the entire economy must follow. 

This is an "iffy" argument which is obviously difficult to test, and no 
satisfactory tests have yet been devised. If the hypothesis is correct, one 
could say that administered pricing arrangements contribute indirectly to 
cost inflation by providing a more permissive atmosphere for wage increases. 

Worker expectations 

It clearly makes a difference whether workers expect substantial wage 
increases to occur every year, or whether they expect wage increases to be 
moderate and intermittent. Worker expectations generate union policies, 
political pressures, and individual demands on employers which may have 
considerable effect on the course of wages. 



9 United Nations, World Economic Survey, 1957, p. 34. 
127 



Little is known about worker attitudes on this point and how they are 
generated. Experience over the recent past must have some influence. If 
wages have been advancing rapidly for a number of years, it is natural 
for workers to project this trend into the future; and if living costs have 
also advanced considerably, they will be all the more insistent on sub- 
stantial wage gains. One of the most difficult features of the present 
situation in many countries is the history of unbroken wage and price 
increases over the past twenty years. There will soon be few workers living 
who can remember a wage cut or even a year in which wages did not 
rise. This has generated a momentum in money-wage movements which is 
much harder to check today than it would have been a decade ago. 

A related matter is worker expectations about union objectives and 
accomplishments. At one extreme, workers may regard their union dues 
as a business investment and judge the union's effectiveness by the mone- 
tary gains which it wins. At the other pole, workers may regard unionism 
as a political and social movement, and may attach primary importance to 
worker participation in management, nationalization of industry, or redis- 
tribution of income through government. Wage bargaining is never un- 
important, but it plays a more central role in some union movements than 
in others. In the years since World War II, for example, it has been less 
significant in Germany than in Britain, and less significant in Britain than 
in the United States. 

The structure of unionism and collective bargaining 

The linkage between wage inflation and collective bargaining is a good 
deal looser than is often assumed in popular discussion. Unionism is not a 
prerequisite for the appearance of wage inflation, nor does the presence of 
unionism guarantee that wage inflation will follow. Unionism does make a 
difference, however, and the characteristics of collective bargaining in a 
particular country can be of substantial importance. Most significant 
among these characteristics are probably the following: 

1. The incidence of union organization. It makes a difference whether 
unionism is strongest in the sectors where conditions are most favorable to 
money wage increases, or whether the contrary is true. In the United 
States, strong textile unionism and weak automobile unionism would pro- 
duce a rather different wage atmosphere than exists at present. 

2. The locus of control over union policy. Aggressive membership 
participation in union government and insecurity of tenure among union 
leaders may be applauded on democratic grounds, but it probably also 
makes for larger wage demands than would occur otherwise. Top union 
officials have better economic information, a longer-range outlook, and 
greater concern with employment and other side-effects of wage increases 

128 



than do union members. To the extent that the leaders can proceed with- 
out direct membership control, they are likely to be more moderate and 
realistic in wage demands and wage settlements. The strongly entrenched 
leadership of British trade unions, for example, has contributed to a less 
frenetic wage policy than prevails in the United States. 

3. Inter-union relations and the scope of collective agreements. The 
labor movements of industrialized countries exhibit wide differences in this 
respect. In some countries, including the United States, local unions typi- 
cally carry on the bargaining with general advice and assistance from 
national headquarters, and the single-company agreement is the general 
rule. In others, including Britain, the collective agreement is typically 
negotiated by national officers and is coextensive with the industry. A few 
countries, including Holland and Sweden, have experimented with still 
more centralized systems in which a central trade union federation strikes 
a master bargain with a national confederation of employers, and this 
provides a general framework for more detailed bargaining in individual 
industries. 

The differences among these collective bargaining systems are not as 
great in practice as they can be made to appear on paper. In American 
industries where employers compete in regional or national markets, the 
national unions try to ensure substantial uniformity in the terms of local 
settlements and the situation often approaches that of industry-wide bar- 
gaining. Conversely, in countries such as Sweden, Norway, and Holland, 
wage decisions are less centralized than they appear to be. The national 
master agreement sets certain guide-lines, but these are later applied and 
modified through separate negotiations in each industry. Similarly, the 
terms agreed on at the industry level may be modified a good deal in their 
application to individual plants and groups of workers. The master agree- 
ment typically ensures a certain minimum increase for every unionized 
worker; but there is a great deal of bargaining up from this minimum, and 
the actual realized increases vary considerably from one plant and in- 
dustry to another. 

There has been considerable discussion of the relation between cen- 
tralization of collective bargaining and the aggressiveness of union wage 
policies. Perhaps the commonest hypothesis is that greater centralization is 
likely to produce more moderate wage demands, demands which are ad- 
justed to national economic necessities and are held within the bounds 
permitted by productivity increases. The rationale for this is that leaders 
of the top union federation are strongly insulated against grassroots opinion 
and pressure, that they have comprehensive economic information and a 
broad outlook on the national economy, and that they are in a position 
to check what might otherwise become a competitive scramble for w r age 
advantage among individual unions. 

129 



This is an interesting and persuasive hypothesis, but one cannot say 
that it has yet been verified by experience. Individual instances can be 
cited. During the late 'forties the labor movements in Britain, Norway, 
Sweden, and Holland, cooperated with their national governments in a 
policy of wage restraint, amounting at times to a virtual prohibition of 
negotiated wage increases. Money wages rose very little in these countries 
from 1947 to 1950, and real wages scarcely at all. In Holland the policy 
of wage restraint continued through the early 'fifties and was sufficiendy 
severe that real wages were actually lower in 1953 than in 1947, and 
labor's share of national income produced declined appreciably. 

One must remember, however, that the circumstances of this period were 
very unusual. The countries concerned were recovering from the physical 
destruction and economic dislocation of a six-year war, and were struggling 
to rebuild their productive capacity and restore their export markets. 
It was thus easier than it normally would be to enlist cooperation of all 
economic groups in a national effort. Incomes of non-wage-earning groups 
were restricted by price controls and heavy profits taxation, which put 
government in a stronger position to ask restraint of wage earners. The 
governments in office were labor governments resting largely on trade 
union support, and union leaders were consequently reluctant to embarrass 
them by opposing their economic policies. 

After the peak of the Korean crisis, restraints on wages and other money 
incomes were loosened in most countries. Price controls and subsidies 
were increasingly abandoned, and the economies moved toward normal 
peacetime operation. From this point on, it is difficult to make a case that 
centralized collective bargaining had a braking effect on the rate of wage 
increase. From 1952 to 1957 money wages rose somewhat more in Britain, 
Holland, and Scandinavia than in Canada and the United States with their 
decentralized bargaining systems. This may have been partly a catching-up 
from the previous period of wage restraint, and partly also a reflection of 
unusually high levels of demand and employment. The West European 
economies have probably operated somewhat closer to capacity during the 
'fifties than those of Canada and the United States, which might well 
account for faster increases in the wage-price level. In any event, collec- 
tive-bargaining procedures and union policies in these countries do not 
seem to have had any clear-cut effect in producing a different behavior of 
the money-wage level than that found in Canada or the United States. 

4. Contract renewals and wage reopenings. In the United States we 
tend to take it for granted that collective agreements must expire and a 
new wage bargain be negotiated every twelve months, but this is in no way 
inevitable. British union agreements are of indefinite duration, though 
annual wage demands have become more common during the nineteen 
fifties than they used to be. Moreover, since there is no contract expiration 

130 



date after which a strike must occur, there is no deadline on the negotia- 
tions, which therefore proceed at a more leisurely pace than in the United 
States. A lag of a year between initial demand and final settlement is not 
uncommon. This slower tempo of wage movements lengthens the wage 
lag during a demand inflation and probably also reduces the likelihood 
of cost-push inflation. 

What to do about it? 

Before discussing remedial measures it should be re-emphasized that we 
are not sure that there is a serious problem of secular inflation, or at any 
rate that the problem is more serious today than in earlier decades. It was 
noted earlier that the rate of price increase in most countries since 1952 
is not at all out of line with the rates in earlier periods of peacetime eco- 
nomic expansion. Price-level increases are also to some extent illusory 
in that quality changes are not adequately represented. One of my col- 
leagues argues that it is quite doubtful whether there has been any genuine 
increase in consumer prices in the United States since 1949. 

The argument is as follows: give a sample of consumers a 1949 Sears, 
Roebuck catalog and a 1959 catalog. Give them also $1,000 each, and 
permit them to order from either catalog. If they choose the 1959 catalog, 
as my friend surmises most of them would do, this must mean that quality 
increases have outweighed price increases and that the price level has not 
risen in any significant sense. 

The question under discussion here, then, should be stated as follows: 
if there is a problem of secular inflation, and if this arises in some measure 
from cost pressure of the wage-push variety (neither of these things being 
certain), what remedial measures might be taken? To the extent that the 
problem exists, it is clearly not amenable to a single once-for-all solution, 
any more than one could hope to "solve" cyclical instability or any other 
deep-rooted structural problem by a single reform. It is sounder strategy 
to think of "weathering down" or encircling the problem by a variety of 
flanking maneuvers. 

The one-shot remedies most frequently proposed for wage inflation do 
not stand up very well under close examination. Some of the commonest 
proposals are: 

(1) Weakening the power of unionism by enforced decentralization — 
for example, by confining collective bargaining to a single company and 
representatives of that company's employees only. This would be a radical 
departure from our past policies in labor relations, since it would mean 
dismemberment of national unions. The proposal is doubtless imprac- 
ticable from a political standpoint, but even if practicable would be unde- 
sirable. In order to eliminate a possible (and in my judgment not really 
serious) adverse effect on the behavior of the money-wage level, we would 

131 



have sacrificed the numerous positive benefits of strong union organization. 
Any decision on such a fundamental matter should clearly be based on a 
comprehensive evaluation of all the effects of unionism — good and bad, 
wage and non-wage, economic and non-economic. My judgment is that an 
evaluation of this sort yields a positive score for national unionism; but it 
is impossible to explain and justify this judgment in the space available here. 

(2) Centralizing collective bargaining in order to "rationalize" wage 
decisions. This recommendation runs directly counter to the one just dis- 
cussed. The argument is that, if bargaining could be centralized along 
Swedish lines, the top federation leaders would be willing and able to 
follow a "responsible" wage policy, i.e. one which does not raise the money 
wage level at an inflationary rate. 

This proposal is also impracticable in the sense that a democratic gov- 
ernment can do little to bring about greater centralization of wage deci- 
sions. The present centralization of decision-making in the Swedish labor 
federation was not planned by anyone, but resulted from gradual internal 
evolution over several decades. The same has been true in other countries. 
In none of the Western democracies has the scope of wage bargains been 
shaped in any large measure by public policy. 

Moreover, the power of top federation leaders to control and moderate 
the pace of wage advance should not be exaggerated. They are leaders, not 
dictators, and political leadership involves survival by judicious responsive- 
ness to pressure. Too tight a control on wages from the center would 
threaten the very structure of the union movement by drying up local 
initiative and membership support. A large measure of wage negotiation 
at the industry and company levels persists beneath the forms of centraliza- 
tion, and these supplementary bargains always work upward from the 
national norms established at the center. It is quite common to find 
Swedish federation leaders prescribing a 3 per cent wage increase in a 
particular year and to see an actual increase of 5 or 6 per cent at the end 
of the wage movement. 

(3) Government intervention in specific price and wage decisions. This 
category embraces a considerable variety of sub-proposals. One of these 
may be termed the "coaxing" approach, under which the government would 
intervene informally in, say, the 1959 basic steel negotiations and persuade 
the parties to accept a modest wage increase and no price increase. An- 
other suggestion is that the government should be given legal power to 
review, or even to veto, proposed wage and price increases. Still another 
suggestion is that the government should announce each year a ceiling 
figure for wage increases — some percentage rate of increase which is judged 
to be noninflationary in view of the expected growth of national output — 
and that proposed increases above this level must be justified to a govern- 
ment agency. 

132 



Proposals of this type appear to be either ineffective or harmful. "Coax- 
ing" can scarcely have much effect unless accompanied by some type of 
sanction. Government review and alteration of wage and price decisions 
runs counter to the rationale of a private economy and could clearly 
hamper efficient use of economic resources. Announcement of a "target" 
figure for wage increases in a particular year implies that all wages can or 
should advance at the same pace over time, which is not the case in a 
dynamic economy. The announced maximum would almost certainly be 
taken as a guaranteed minimum increase throughout the economy, and 
bargaining would proceed upward from there. 

(4) Controlling wage inflation through monetary policy. This is the 
commonest of all proposals in this area, and is often assumed to provide a 
conclusive answer. It is held that price increases resulting from a wage- 
push simply demonstrate that the monetary authorities are not doing then- 
job and that increases can be avoided by a tougher line of policy. If an 
effort is made to push up wages faster than the warranted rate, the mone- 
tary authorities should withhold the additional working capital needed to 
cover the wage increases. This will compel a reduction of output and 
employment, stiffen employer resistance to wage increases, and bring the 
inflationary process to an end. After this has been done several times, the 
unions may "get wise" and refrain from demanding excessive increases 
in the future. 

Space does not permit a thorough evaluation of this line of argument, 
but we may note several deficiencies in it. There are obvious technical 
difficulties: first, of ascertaining that a cost push is in process and judging 
its approximate size; and second, of adjusting money supply to just the 
right extent in face of the notorious instability of velocity noted earlier in 
this paper. More serious, the policy involves a break in the growth trend 
of the economy and loss of output over a considerable period. (Even in the 
mild United States recessions of 1949-50, 1953-54, and 1957-58 it was 
18 to 24 months before the previous output peak was regained, and still 
longer before the economy was back on its long-term growth path.) To 
the extent that such recessions involve a reduction in capital investment, 
the subsequent rate of increase in national output is also reduced, and the 
long-term growth path is lower than it might have been. It is quite possible 
that the depressing effect of a recession on output growth may be greater 
than the depressing effect on money-wage increases, in which case the 
tight-money policy would turn out to have been inflationary rather than 
the reverse. 

The dampening effect on money-wage increases, moreover, is by no 
means certain. Unemployment per se cannot be counted on as very effec- 
tive in an economy of bargained wage rates. The reduction of sales and 
profits during a recession, with the consequent increase in employers' in- 

133 



centive to resist wage demands, is a more powerful force. But it might 
require a substantial cut in output extending over a considerable period 
to really break the momentum of an upward wage movement. The United 
States recession of 1957-58 was not large enough to do this, and brought 
only a slight drop in wage increases in 1958 compared with 1957. 

This line of policy thus appears to be almost a counsel of despair. 
Deliberate creation of unemployment is justifiable only after every alter- 
native has been exhausted, and then only if the costs of reduced output and 
slower economic growth are considered less important than the costs of a 
rising price level. 

What, then, are some of the other approaches which one might take to 
the problem? The most hopeful lines of approach fall under three headings: 

(1) Raising the rate of growth in national output. It is curious that 
economists tend to take this as a datum and to regard the money-wage 
level as the variable which must be adjusted. It is usually argued that if 
the rate of output growth is 3 per cent, and if workers and union leaders 
demand wage increases of 5 per cent, the problem is to persuade them to 
desist and settle for 3 per cent anyway. Suppose that one turned the prob- 
lem around, took the 5 per cent annual wage increase as a datum, and 
asked how we might be able to raise the rate of output growth to this 
level. The rate of increase in output is certainly not invariant. The indi- 
cations are that it has increased materially over the past two or three 
decades. Perhaps it could be further increased by heavier investment in 
scientific and technological research, by policies designed to encourage 
business investment in plant and equipment, and by other methods. 

Putting the problem in these terms leads to different policy conclusions 
from the alternative approach. It leads to an emphasis on a high growth 
rate for the economy, a high level of investment, and continuous capacity 
operation. It leads, in short, toward a positive monetary policy oriented 
toward economic expansion rather than a restrictive policy designed to 
punish excessive wage demands through periodic unemployment. This 
positive policy should not mean open courting of inflation, and it ad- 
mittedly contains some danger of encouraging excessive wage increases. 
One must hope, however, to achieve an acceleration of productivity growth 
which will more than offset any intensification of wage demands. Nor will 
wage demands necessarily be intensified if some of the other measures 
suggested below are taken. 

The importance of looking at the output side of the picture as well as 
the wage side is underlined by the events of 1956-57 in the United States. 
The Consumer Price Index rose by 5 per cent between 1955 and 1957, and 
this was attributed by many to excessive wage increases. Actually, the 
increase in money wages during 1956 and 1957 was very closely in line 
with the rate of increase in preceding years. The rate of output growth 

134 



during 1956 and 1957 was abnormally low, however, with the necessary 
result of a rising price level. The significant question to ask about these 
years is not "Why did wages rise so much?" but rather "Why did output 
increase so little?" 

(2) Increasing public understanding of economic affairs. This is an 
obvious and even trite suggestion, but one whose long-run importance 
should not be discounted. After a century of economics teaching in the 
United States, popular discussion of wage issues is still studded with obvious 
fallacies concerning the interrelations of wages, profits, income distribution, 
prices, and output. The level of economic literacy in collective bargaining 
is noticeably lower in the United States than in Britain or Scandinavia. To 
say that this situation cannot be improved is to abandon hope of rationality 
in economic affairs. Informed self -discipline by economic groups is the 
alternative to state compulsion, and better economic data and economic 
analysis are an essential basis for this. Education in the economics of 
wages cannot come in any great measure from the parties at interest, 
whose motives are suspect and whose "informational" material typically 
reflects self-interested pleading. The lead must be taken by disinterested 
economists in universities and research agencies, who until now have failed 
to make any marked impact on the general public. Popular, literate, eco- 
nomically informed writing for a mass audience needs to be made as 
respectable in this country as it already is abroad. Credit should go also 
to the small band of economists who are trying with some success to 
improve the economic content of history and social studies courses in 
the high schools. 

(3) Institutional reform. I pointed out earlier that the susceptibility 
of an economy to wage inflation depends on a variety of structural char- 
acteristics. These are often deep-rooted and incapable of rapid change 
through government action. Over a period of decades, however, economic 
institutions do change and the rate of change can often be accelerated by 
conscious forethought. It is important, therefore, to be clear about the 
direction of movement which would be desirable. 

Desirable directions of movement in the United States at this time 
would include: 

a. Improvement of labor markets and rationalizing of labor mobility 
both within localities and between localities. This is desirable for many 
reasons. In the present context, it would have the consequence of flattening 
labor supply curves, reducing the chances of labor supply bottlenecks at 
growth points of the economy, and consequently reducing the size of wage 
increases needed to reallocate the labor force in new directions. 

b. Strengthening of competitive forces in product markets and discour- 
agement of open or tacit price agreement. Henry Simons, that eloquent 
advocate of a liberal economy, was wrong in asserting that a unionized 

135 



industry is "doomed to full cartelization." Witness textiles, clothing, shoes, 
food processing, and many other branches of industry in which unioniza- 
tion has coexisted with vigorous price competition. A national union does 
not produce a cartel, but it will ride the coat-tails of a cartel or an 
oligopoly group which has come into existence for other reasons. It may 
well demand wage increases larger than could be obtained under com- 
petitive conditions, and large enough to have inflationary consequences. 
Vigorous enforcement of competition in product markets thus strikes in- 
directly at a possible source of undue wage pressure. 

c. Strengthening employer solidarity in collective bargaining. This may 
seem logically to run counter to the previous proposals, but pragmatically 
it need not do so. There is no reason why employers cannot present a 
united front to the union in collective bargaining while continuing to com- 
pete actively on other fronts. In retailing, service industries, and some 
small-scale manufacturing industries, growth of national unions with no 
corresponding organization on the employer side has produced for the 
time being an imbalance of bargaining power in favor of the union. Even 
the large and powerful automobile companies find themselves at a disad- 
vantage when the union can pick them off one at a time. Initiative in 
rectifying such imbalances of bargaining power must come from employers 
themselves, but it could at least be made clear — by fresh legislation if 
necessary — that joint bargaining by employers does not contravene the 
antitrust laws. The alternative approach of breaking down the national 
unions to company-size units is neither feasible nor desirable for reasons 
already given. 

d. Building lags into wage adjustments. The hurried, nervous procession 
of annual wage "rounds" which we have experienced since World War II 
may be unavoidable in an economy undergoing continuous inflation; but it 
should not be necessary in a country committed to a reasonable measure of 
price stability. We can doubtiess not go all the way to the British pattern 
of agreements without a terminal date; but three-year and even five-year 
agreements are certainly to be encouraged. There would have to be pro- 
vision for annual reopening of the wage issue; but this does not compel a, 
major wage push to the same extent as actual expiration of an annual 
agreement. Salary schedules for civil servants, college teachers, and other 
white-collar people, are typically raised only every two or three years, and 
since this is customary it is generally accepted by those concerned. There 
seems no reason why manual workers, whose conditions are becoming 
more like those of white-collar people in most respects, should not also 
adjust to a more leisurely pattern of advance in wage schedules. 



136 



6 




Policy problems: 

Choices and proposals 



John T. Dunlop 



In the United States during the period 1953-58 gross hourly earnings for 
production workers increased 4.1 per cent a year in manufacturing; the 
Consumer Price Index rose 1.6 per cent a year; wholesale prices (other 
than farm products and foods) rose 2.1 per cent a year, and output per 
man-hour in the private economy increased almost 3 per cent a year. Total 
compensation including fringe payments probably increased 4.5 per cent a 
year in this period. The prospects of a long-term or secular average rise 
in prices of 1.5 to 2.0 per cent a year is one way of stating the central 
policy problem. The purpose of this paper is to state and to appraise, from 
one man's point of view, the major policy alternatives and choices for the 
scrutiny of debate and discussion. 



John T. Dunlop, Professor of Economics at Harvard, has served in various govern- 
mental capacities, including: Staff Member, National War Labor Board (1943-45); 
Consultant, Office of Economic Stabilization and Office of War Mobilization and 
Reconversion (1945-47); Member, Atomic Energy Labor Relations Panel (1948- 
1953); Public Member, Wage Stabilization Board (1950-1952); Member, Emergency 
Board, Conductors Case (1955). He served as Impartial Chairman, National Joint 
Board for Settlement of Jurisdictional Disputes in the Building and Construction 
Industry (1948-1957). 

Professor Dunlop is author of Wage Determination Under Trade Unions; Col- 
lective Bargaining: Principles and Cases; The Theory of Wage Determination, 
(Editor); Industrial Relations Systems; and other books and articles. 

137 



The policy issues raised by the experience since 1953 are often distin- 
guished from the classic inflations related to World War II and its aftermath 
and to the sudden distortion of expectations and world raw-materials 
prices associated with Korea. While all periods of inflation may be 
treated as a whole at some level of generality, and while the events of 
1939-52 no doubt decisively shaped many features of the ensuing years, a 
relatively new species of inflation — the long-run, secular, creepy kind — 
is of current concern. 



Types of policy prescriptions 

A year ago in England I also found intense interest and discussion over 
long-term inflation. 1 Average earnings had increased 7.5 per cent and 
prices had risen almost 4 per cent a year on the average in the preceding 
five years. The British discussions suggested four different lines of diag- 
nosis by the local economic doctors. Each view of the source of the 
malady and each explanation of the inflationary symptoms carried its own 
policy prescription. In considering the experience and discussion in the 
United States since 1953, with minor variations, the same four groups of 
diagnoses and prescriptions may be distinguished. 

(1) According to the first group of economic doctors, the central dif- 
ficulty is an excess of demand. The unwarranted expansion in credit and 
in the velocity of circulation under a variety of monetary authorities 
stimulates a rise in wage rates and prices. The disease is of the circulatory 
system, an excessive secretion of the monetary gland. Secondary complica- 
tions soon set in with this well-known ailment, and the British patient 
suffers an acute loss of sterling, while in the case of the United States there 
is even concern over seepage from Fort Knox. 

"On my conception, up to recently the rise of wages has usually been 
the result, not the cause, of the general inflation." 2 

". . . the 1955-57 inflation was in fact basically similar to inflations of 
the past ... the role of costs in this inflationary episode has been greatly 
exaggerated." 3 

The cure prescribed is a tight monetary and fiscal policy to eliminate 
excess demand. Some of the doctors, regarded as less sound by those who 
pride themselves as being most sound, are worried that the amount of this 
medicine required to cure may be almost as bad for the patient as the 



1 See for instance, the symposium on wages policy, Scottish Journal of Political 
Economy, June, 1958; Council on Prices, Productivity and Income, First Report, 
February, 1958; Second Report, August, 1958. 

2 Lionel Robbins, "Thoughts on the Crises," Lloyds Bank Review, April, 1958, p. 23. 

3 Richard T. Selden, "Cost-Push versus Demand-Pull Inflation, 1955-57," Journal of 
Political Economy, February, 1959, pp. 1-2. 

138 



illness itself. The medicine may damage output and real investment essen- 
tial to the long-term growth and recovery of the patient. 

(2) The central difficulty, says another group of diagnosticians, lies 
deeper in the economic body. "The removal of excess demand does not 
necessarily restore equilibrium ..." 4 or economic health. The real problem 
is that money rewards rise faster than productivity. The malady is wage 
costitis. The old-fashioned remedy was a complete rest — idleness for five 
to ten per cent of the work force. But most modern physicians regard this 
cure as too drastic; the patients these days simply will not take such 
medicine. Modern doctors are a little uncertain what to prescribe for wage 
costitis. There is the sugar pill of advice to unions and managements to 
take sweet moderation three times daily before negotiations. A number of 
other research workers have been discovering the bargaining mechanism, a 
part of the economic anatomy hitherto left to a little-regarded fringe of the 
profession. Some have associated the disease of inflation with unco- 
ordinated or sectional bargaining, and they propose an operation, of 
Swedish or Dutch origins, to graft on centralized bargaining. Others have 
associated the malady with enlarged bargaining units, and they propose 
that the anti-trust knife cut out industry-wide bargaining and sever the 
nerves that connect local unions and national unions. It is understandable 
that with such conflicting advice the patient has stoutly resisted submitting 
his bargaining mechanism to any operation. 

(3) The central difficulty, says another group of economic doctors, is 
psychological. The malady is of the economic mind or will. The patient 
wants too much of too many good things; the combination is beyond its 
means. The patient refuses to face reality and stern choices. It wants a 
continuing high degree of employment, full utilization of capacity to meet 
demands from home and abroad, increased leisure and shorter working 
weeks, the full continuation of the era of human relations and labor peace 
which has been adopted in an increasing number of enterprises since the 
end of World War II, the absence of all direct government controls over 
production, prices and wages, no more stringent or detailed fiscal or mone- 
tary controls, and only an occasional dose of monetary aspirin when 
critically ill. In the British case the patient wishes to continue to play the 
role of banker for a large part of the world on small reserves, and the 
inflation morality of a banker must be more correct than others. In the 
case of the United States, the patient cannot bring himself to abandon 
the flesh pots and face the stern discipline of economic competition with 
the East. The widespread view that a high degree of all desirable goodies 
can be simultaneously achieved is a neurosis. There is nothing structurally 



4 F. A. Burchardt, "Cost Inflation," Bulletin of the Oxford University Institute of 
Statistics, November, 1957, p. 327. 

139 



wrong with the patient; it wants too much of too many conflicting objec- 
tives. It fails to understand the elementary concept of opportunity costs. 
The stern prescription is simply to choose a more limited group of objec- 
tives and to arrange a clearer scale of priorities: to take more unemploy- 
ment, or more labor strife, or more specific fiscal and monetary controls 
or more direct controls over wages, prices and production, or reshape one's 
responsibilities abroad to accommodate to the resources. 

(4) The central difficulty lies in temporary or permanent structural 
defects in the economy, says another group of economic doctors. In the 
British economy, increasing levels of investment in machinery and equip- 
ment have shown relatively small results in productivity and output per 
worker. Coal mining and railroads are older industries, and a great deal 
of capital is required to overcome the difficulties of a misspent youth or the 
natural failings of old age. It is doubtful if a high rate of increase in pro- 
ductivity can ever be restored in these industries, but they are indispensable 
to the economic body. A less pessimistic view would hold that the capital 
has been flowing in Great Britain into an infrastructure of power, trans- 
portation and services which have not shown full results in these five years, 
but the increasing investment has been for a brighter long-run future. In 
the case of the United States the temporary or more permanent structural 
defects include such items as the failure of agriculture to provide lower 
prices, the shifting composition of output toward government activities and 
the service industries in which productivity increases appear to be lower or 
unmeasurable, the inflation-generating role of oligopolies and the rise of 
strong unions in such sectors. In the view of this school of doctors the 
malady is a temporary or more permanent defect in organic structure, and 
any cure is always slow and painful. 

The preceding brief survey of major points of view respecting secular 
inflation emphasizes that analysis and policy prescription are usually 
closely linked. 

The community's priorities and preferences 

In his inaugural lecture as professor of political economy in Cambridge 
University, Professor J. E. Meade said: 

I personally find little difficulty in reaching the conclusion that inflation as well 
as deflation is an evil which deserves a very real effort to avoid. But let us be 
very clear that there are worse evils than inflation. The stagnation of the 
1930's is a much worse evil than the inflation of the 1950's. We must give very 
careful thought to the question whether a serious attempt to prevent a con- 
tinuation of the inflation of money prices would lead to a cessation of growth 
of economic output and to a rise in the general level of unemployment. 6 



5 The Control of Inflation, Cambridge, England, Cambridge University Press, 1958, 
p. 11. 

140 



What are the relative preferences of the community for price stability 
and for other objectives which may have to be given up in order to achieve 
price stability? What are the social opportunity costs of stability? What 
is the price of a lesser degree of price inflation? All economists should be 
cautioned against imposing their own preferences upon the community. 
What the community ought to prefer by some standard and the priorities 
it does in fact establish are not to be confused. What is the distribution of 
priorities among different segments of the community? Price stability can 
only be achieved at a price or at the cost of other values forsaken. Before 
anyone gives price stability the top policy priority, it is essential to explore 
some of the costs of stability. 6 

The price of stability 

( 1 ) How high a price is the community willing to pay for price stability 
in terms of more labor strife? A lesser rate of wage increase — and a lesser 
rate of price increase — probably requires more strikes. If managements are 
to secure settlements on the average below a figure of 4 or 5 per cent per 
year, or 8 to 10 cents an hour in manufacturing, there will probably be 
more strikes over the money package at contract renewal times. But is 
there much disposition on the part of managements to take a strike over 
another 2 or 3 cents when profits are high, when relations have become 
friendly, when a strike may cause an intensive search for alternative sources 
of supply and substitutes on the part of customers, and when the friendly 
cooperation of the union and work force is vital to the introduction of 
new machinery and other means of raising productivity? It is not the 
preachments on the part of national employers' associations against "in- 
flationary" wage increases that counts, nor the urging on another firm to 
take the lead ("Let's you and the union fight"), nor the smooth and 
catchy speech of business leaders printed in attractive form and distributed 
to stockholders, the press and educators. It is what the firm actually does 
at five minutes before midnight when it must balance going a little higher 
on wage rates against taking a strike with the related complaints of cus- 
tomers and potential loss of accounts. It is on such occasions that the 
price of stability is tested. 

(2) How high a price is the community willing to pay for stability in 
terms of unemployment? It is possible to enhance wage and price stability 
by creating a higher average level of unemployment. In the six-year period 
1953-58 unemployment averaged 4.7 per cent of the civilian labor force, 
4.3 per cent for the period 1953-57. In 1958 the figure averaged 6.8 per 



6 See, John T. Dunlop, The Secular Outlook, Wages and Prices, Institute of Industrial 
Relations, University of California, 1957, pp. 14-17. 

141 



cent. These figures are considerably above unemployment rates in Great 
Britain and most of Western Europe. Is the American community prepared 
to increase these figures to achieve more stability? Or is the community's 
desire for goods, its fear of unemployment and its concern with compara- 
tive production records of the Soviet bloc such that it would be unwilling to 
trade more unemployment for additional price stability? It is easy to want 
both higher levels of employment and more price stability. But in the hard 
choices the community makes in the market and through its elected repre- 
sentatives it reveals the price it is willing to pay for stability in terms of 
unemployment. It is well to remember that the American community 
already tolerates a higher level of unemployment (as a percentage of the 
civilian labor force) than most northern and western European countries. 

(3) How high a price is the community willing to pay for price stability 
in terms of a wider range, more detailed and more flexible government 
controls? The community comes to see that any system of controls over 
inflation has its limitations and can be pushed only so far, and that each 
control device has some drawbacks. While direct controls are likely to be 
most effective, at least for short periods, in stopping upward movements, 
the community most distrusts these devices. Every schoolboy can tell of 
the evils of bureaucracy and the effects of direct controls on distorting the 
most effective allocation of resources. But general monetary and fiscal 
controls also have their limitations. Small businesses, home builders, farm- 
ers, state and municipal governments and other types of borrowers may be 
relatively disadvantaged by stringent monetary controls. A larger pro- 
portion of capital may go to established borrowers or to those who can 
finance from within than may prove desirable to the community. A more 
flexible and detailed system of fiscal controls with more sensitive and 
variable tax rates and allowances has not been devised on an acceptable 
basis. Such fiscal controls would seem to require modification in accepted 
views of the process of exercising effective taxing powers. In the willing- 
ness to accept further controls — direct, monetary or fiscal — is to be seen 
the price the community is willing to pay for stability. 

We need to be quite explicit about our own individual priorities or 
preferences for price stability relative to other objectives and also to be 
quite clear about our estimates of the preferences of the various groups 
and the aggregate which comprise the American community. We can 
avoid much needless debate by being clear on these matters at the out- 
set. What price is each of us prepared to pay for increased stability in 
terms of more labor strife, more unemployment or more detailed and 
stringent government controls? While estimates of the community prefer- 
ences will vary widely among us, I submit there is not much evidence to 
indicate the community is now willing to barter for more stability. 

142 



I trust we are involved in no sentimental condemnation of inflation 
without also measuring the opportunity costs or the price of the measures 
required to induce greater price stability. Some degree of price stability 
is an affirmative value, and some degree of inflation is an evil. This is 
not an issue, but rather the policy choice is how much are we willing to 
pay, what are we willing to give up for more stability, and what do we get 
for more inflation. This is a tough issue and we need a tough-minded 
approach to these hard choices. 

I respect the relative priorities which each person selects with internal 
consistency among the four sets of objectives noted above: (1) stability 
or inflation, (2) labor peace or industrial strife, (3) employment or un- 
employment, and (4) government controls or free markets. 

Personal preferences 

There is probably no standard by which to adjudge one consistent 
combination of preferences superior to another. There is one preference 
of mine, however, which should be made explicit, namely, that the Ameri- 
can community has been paying too high a price for the degree of stability 
achieved in terms of unemployment, loss of production and loss of relative 
position with the Soviet bloc. In my preference scales, we can no longer 
afford, even at the cost of more inflation, the average level of unemploy- 
ment of the past five years. 

Barbara Ward has said, ". . . the technique of checking production in 
order to stop inflation — the most drastic and dangerous 'cure' conceivable 
in the face of Russia's rising production — does not cure at all. Inflation 
goes on. New thinking is thus in order." 7 

Professor Sumner Slichter has stated the relation between inflation and 
economic growth in these words: 

The way to accelerate the growth of the economy is to stimulate the demand 
for goods. When demand presses hard upon productive capacity, that capacity 
is used and industry takes steps to increase its capacity. Pressure of demand 
upon capacity tends to produce creeping inflation, but this pressure is necessary 
to attain the maximum rate of growth. 8 

In my scale of preferences the United States requires a much higher rate 
of growth and capital formation in the next five years than we have had 
in the last five, and it is imperative to our national interests that we use 
our labor force to a higher degree. This will further complicate the in- 
flation problem, and I am glad for the present discussion, for the concern 
with inflation has no doubt been a major factor retarding public policies 



7 "Now the Challenge of An Economic Sputnik," The New York Times Magazine, 
February 8, 1959, p. 66. 

8 "The American Economy — Current Trends, Problems and Prospects, January and 
Early February, 1959," Prepared for Ninon Keizai Shimbun of Tokyo. 

143 



designed to create a higher rate of utilization of resources and to create new 
capacity. Whether you accept these personal priorities or not, we should be 
able to agree upon the policy alternatives involved in restricting price 
(and wage) inflation. When we come to express choices among the alter- 
natives, we shall need to make explicit our relative priorities for stability, 
labor peace, full employment, a rapid rate of growth and freedom from 
additional government controls. 

There will be those who will resist making these choices, feeling that 
"new thinking," to use Barbara Ward's phrase, is in order. The policy 
decisions are to be sought less in choices within the existing framework than 
in changes in the private and public institutions which make the decisions 
on wages and prices in the economy. The section which immediately 
follows considers some proposals for changing the institutional framework. 
These particular ideas seem to me to offer no alternative, simply because 
they are not likely to produce the results on wage and price formation 
which their sponsors claim. But they have received widespread attention. 

The wage bargaining institutions 

Internal union reform 

There is an oft repeated view that strong upward pressure from wages 
is attributable in part to an alleged lack of "democratic procedures" within 
labor organizations, and that reform of internal union government will 
make a significant contribution to the control of inflation. A variety of 
internal union reform legislation, including provision for government-con- 
ducted strike votes, has been seriously argued for on these grounds. The 
premise is that the leadership has been pushing for wage increases, even to 
the point of strikes, which the rank-and-file would otherwise forgo. In- 
ternal union reform issues should stand on their own merits in my view, 
unrelated to the inflation discussion; union reform in general is not before 
us. But even Professor Arthur F. Burns, who apparently agrees to the 
separation of these questions, seems to regard internal labor reform legis- 
lation as having an indirect impact on prices. In a chapter on "Policies 
for Coping with Inflation," he says: 

The least we can do with regard to trade unions is to subject their finances, as 
well as the election of their officials, to standards defined by law. Such legis- 
lation would of itself have no effect on what happens at the bargaining table; 
but it should help to remind the leaders of our trade unions that unless they 
practice greater restraint and foresight, the government may need to take 
drastic steps to curb their power to push up costs and prices. 9 



9 Prosperity Without Inflation, New York, Fordham University Press 1957, p. 84. 
A similar statement by Professor Burns appears in United States Monetary Policy: 
Its Contribution to Prosperity Without Inflation, The American Assembly, Columbia 
University, New York, 1958, p. 215. 

144 



The attempt to support proposals for internal reform of unions by resort 
to the inflation argument reflects a serious misunderstanding of the actual 
relationships between union officers and members. A leadership more 
responsive to the immediate preferences of workers would no doubt on 
balance result in larger wage increases and more strikes. This view is sup- 
ported by the experience with bargaining under union rivalry and weak 
leadership and by the fact that the bargaining process necessarily requires 
that in most settlements workers accept the recommendations of the leader- 
ship. The experience with right-to-work legislation is further evidence. It 
appears that some degree of weakening in the status of the union leadership 
often leads to more militancy and the processing of more questionable 
grievances. 10 The line of argument is misinformed which holds that union 
power pushing up wages is derived from the role of leaders and that legis- 
lation reducing the power of union leaders or providing for more detailed 
participation by union members in decisions of the union will curb wage 
increases. It reflects a lack of understanding of union government and 
bargaining procedures. The historical experience of our unions with the 
referendum and the factors which led to the rise of the national executive 
demonstrates that our present union government was not capriciously 
achieved. 11 It is time that the case for needed internal labor reform be 
separated from the inflation problem. Indeed, it is likely that we shall 
have to pay for internal labor reform in the form of some further wage 
increases as officers are made more timid or insecure and their powers to 
deal with dissidents are limited. 12 

Area of bargaining 

The view is often expressed these days, by serious students of economics 
who are without experience in labor-management relations, that the power 
of unions to raise wages should be curbed by legislation designed to cir- 
cumscribe the area of bargaining or to limit the methods of conflict or 
pressure. These proposals include the further application of the antitrust 
laws to unions and wage setting, the further restriction of boycotts and 
secondary action and further limitations on strikes and picketing. 

Concretely, prohibition of violence and coercion, especially of picketing, pro- 
tection of those willing to work, prohibition of closed shop agreements and 



10 Frederic Meyers, Right to Work in Practice, New York, The Fund for the Re- 
public, 1959. 

11 Lloyd Ulman, The Rise of the National Trade Union, Cambridge, Harvard Uni- 
versity Press, 1955, pp. 203-301. 

12 Limitations on the powers of the international union to place locals under trustee- 
ship are likely to reduce the intervention of the international in local negotiations 
when the local is striking for increases regarded as excessive by the international. 
The case of the International Union of Operating Engineers and its Philadelphia local 
is instructive on this point. 

145 



application of the Sherman Act to trade unions, or, to put it briefly, a mild 
stiffening of the Taft-Hartley Act, may be sufficient to bring about the de- 
sired results. 13 

This is not the forum in which to discuss public policy on the legal 
framework of labor-management relations. But the issues cannot be 
avoided insofar as such proposals are urged to make a serious contribution 
toward holding down secular price increases, such as took place in the 
period 1953-58. I believe serious economists who support these views 
have an obligation to develop the details of such proposals and to submit 
them to scrutiny if they are to continue to urge such policies. It is my 
considered judgment that such policies are impractical, unadministrable and 
would not produce the intended results on secular inflation. I have delib- 
erately not used the argument that such proposals are politically unpopular, 
for popularity is fickle. 

More specifically, the proposals to break up industry-wide bargaining 
usually refer to nation-wide settlements, such as steel, automobiles, coal 
and railroads. But the argument in support of the break-up of negotiating 
districts by legislation depends on market-wide wage setting, local, regional, 
or national. The proposal to break up wage negotiations that are product- 
market wide would involve a vast paper reconstruction of bargaining 
mechanisms which could not succeed basically because it fails to recognize 
that wage interdependencies in the labor market, and the scope of uniform- 
ity of wage rates or changes in wage rates, derive basically from product- 
market interdependencies among enterprises. Market-wide wage setting is 
involved in many local or regionally oriented industries — garment, con- 
struction, hotels, printing, trucking, etc. In theoretical terms union bar- 
gaining power may be greater in some of these sectors than in national 
market-wide bargaining. Market-wide wage setting — either in a single wage 
conference or by wage leaders setting a pattern generally followed within a 
wage contour — arises not because legislation has conferred status or power 
on unions but because of product market inter-relations. Legislation did 
not create market-wide wage setting and legislation cannot take it away. 

We need in this field, as in others, more appreciation of what can be 
accomplished by law and what is beyond the method of legal enactment. 
We need to discern when legislation results only in the dead letter, the 
loophole, evasion, administrative blackmail, the technical quagmire, per- 
petual litigation, and the lawyer at the elbow of every decision-maker. 
This is no condemnation of particular proposed legislation, but in some 



13 Gottfried Haberler, "Wage Policy, Employment, and Economic Stability" in The 
Impact of the Labor Union, David McCord Wright, ed., New York, Harcourt, Brace 
and Company, Inc., 1951, p. 61. Also see, E. H. Chamberlin, Labor Unions and 
Public Policy, Washington, D. C, American Enterprise Association, 1958, pp. 29, 45. 

146 



areas of labor-management legislation, such as some areas of picketing and 
secondary boycotts, we have already reached saturated confusion. 

Instead of splitting up areas of uniform wage setting, a more likely 
development is the transformation of wage leadership — under which most 
enterprises in the wage contour follow the wage adjustments of the leaders 
— into more formal arrangements for market-wide bargaining. In the basic 
steel industry the single wage leadership of the United States Steel Cor- 
poration has been gradually replaced in the past decade by a formal 
negotiating committee representing a number of the major producers. In 
the automobile industry the negotiations of 1958 saw a much closer rela- 
tionship among the big three producers than in previous years, although 
joint negotiations have not yet emerged. The airlines industry in 1958 saw 
an agreement among carriers to share gains attributable to a strike with 
the struck company, in limited circumstances. The growth of more formal 
bargaining arrangements on a multi-company basis may be expected in 
meat packing, aluminum and other industries in which wage leadership 
and pattern followers has prevailed. 

Managements increasingly believe that their bargaining position may be 
enhanced and settlements can be made at lower figures if they negotiate as 
a group, particularly in wage contours where they have previously followed 
a pattern-making settlement. This is the likely direction of development in 
bargaining institutions rather than any artificial or paper fractionalizing 
of bargaining. It is difficult to predict the magnitude of impact of such 
developments on the size of contract settlements, but it is possible that 
contract changes may be more modest and that the larger scope of bar- 
gaining may more explicitly consider the consequences of settlements on 
the larger community. 

Preachments 

One of the most persistent ways in which government spokesmen have 
sought to influence the results of bargaining institutions is through gra- 
tuitous advice and general preachments to the parties. The following 
excerpts from the Economic Report of the President are illustrative: 

Of particular importance in a prosperous economy is the responsibility of 
leaders of business and labor to reach agreements on wages and other labor 
benefits that are consistent with productivity prospects and with the main- 
tenance of a stable dollar. 14 

Leaders of labor unions, in view of the great power lodged in their hands, have 
a particularly critical role to play. Their economic actions must reflect aware- 
ness that stability of prices is an essential condition of sustainable economic 
growth and that the only road to greater material well-being for the Nation 



14 Transmitted to the Congress, January 23, 1957, p. 12. 
147 



lies in the fullest possible realization of our productivity potential . . . Self- 
discipline and restraint are essential if agreements consistent with a reasonable 
stability of prices are to be reached within the framework of the free competi- 
tive institutions on which we rely heavily for the improvement of our na- 
tional welfare . . . 15 

Preachments from this sort of text do not give much promise of driving 
out the devil of inflation. Professor Arthur F. Burns has well said: 

Official appeals for restraint in wage and price adjustments may be salutary, 
but experience suggests that it would be unwise under ordinary circumstances 
to expect a broad response to exhortation. 16 

A command to halt wage and price increases spoken by the President 
has no more effect on the tides of inflation than the words of King Canute 
confronting the rising waters of the English Channel. 

Despite this hard-headed judgment, it seems to me that the full potential 
of the leadership of the federal government has never been used persistently 
and imaginatively to shape decisions by private parties on wages and prices 
or to influence the climate of ideas within which such decisions are made. 
Over long periods, it is no doubt substantially true, that as Keynes con- 
cluded in the General Theory, ". . . it is ideas, not vested interests, which 
are dangerous for good or evil." If the federal government expects to 
influence the ideas of the parties to collective bargaining, it must leave the 
level of repetitive platitudes and generalities and meet with labor and man- 
agement representatives regularly to discuss and debate in free exchange 
and with detailed statistics the economic setting and outlook in which wage 
and price decisions are made. More specifically, the following suggestion 
might be tried. 

After the Economic Report of the President has been transmitted to the 
Congress and there have been hearings before the Joint Congressional 
Committee each year, the Secretary of Labor might convene each year in 
the early spring a three-day conference with leading representatives of labor 
and management. The Chairman of the Council of Economic Advisers, the 
Secretary of the Treasury and other government officials should present 
their detail analysis of the short-term and the long-term economic outlook. 
Representatives of management and labor should be given the opportunity 
to discuss these views of the Administration and to present materials and 
judgments of their own. In addition to formal sessions there might well be 
informal off-the-record periods of free give-and-take. 

The purpose of these annual discussions should be to develop a con- 
sensus of opinion, insofar as possible, or to narrow the range of views 
concerning the major problems confronting the economy as a whole and 



15 Transmitted to the Congress, January 20, 1959, pp. 5-6. 
18 hoc. Cit. t p. 75. 



148 



the expectations of the short-term and longer-term business outlook by 
principal sectors. These discussions would not be negotiations nor should 
they be designed to prejudice any particular contract negotiations. But 
the government would help to sketch the economic problems and climate; 
the interchange would benefit all three groups. The fundamental point is 
that imaginative government leadership should press beyond annual cau- 
tions and preachments to more direct exchange of ideas and information 
in a society of free men and free collective bargaining, and particularly in 
a society in which the level of general education and the detail and the 
quality of statistical and economic information has been growing rapidly 
in the past generation. 

Strategic sectors of secular inflation 

In recent years there have been many studies of the mechanics of inflation. 
In the main these models have been concerned with the short-run relations 
between costs and prices, or among consumption, investment and income. 17 
These aggregate models have been built with varying assumptions con- 
cerning the decision making of households, enterprises and governments 
and with varying lengths of lags in reactions within a system. They have 
included various wage functions. 

In passing, I should like to state briefly my own views about appropriate 
wage functions in such models of inflation. Wage determination most fruit- 
fully should be divided into two parts — the determination of the wage rate 
schedule and the determination of the wage drift, defined as the change in 
hourly earnings arising from other than changes in scheduled wage rates. 
Changes in the wage drift are a function of levels of output, unemployment, 
rates of technical change and of premium hours. Changes in scheduled 
wage rates are primarily a function of profits and consumers prices. Even 
such general statements have implications for policy; for instance, high 
profits can be expected to attract wage-rate increases. It is as much in the 
nature of a labor organization to seek a part of high profits as it is for a 
cat to stalk birds. 

The threshold to inflation 

Generalized propositions and models of inflation suggest that policy 
might well be oriented toward raising the critical level of employment (or 
unemployment) at which wage and price increases accelerate as the system 



17 See, James S. Duesenberry, Business Cycles and Economic Growth, New York, 
McGraw-Hill, 1958. 

149 



moves from low levels to high levels of business activity. What can be 
done to raise the threshold to inflation? What can be done to lower the 
degree of inflationary response to high levels of activity? 

In 1954, under the auspices of the International Economic Association, 
a number of economists met at Seelisberg, Switzerland, and in the course of 
their discussion of wages they considered the factors determining the in- 
flation potential of a system. The propositions developed in that inter- 
national discussion provide a useful starting point for policy, and they may 
be briefly summarized as follows: 18 

1. The less unbalance there is in the structure of production and the 
more evenly spread the expansion of demand, the higher the level of em- 
ployment without inflation. 

2. The greater the increase in productivity and the more evenly spread 
the increases in productivity among sectors, the higher the level of em- 
ployment without inflation. 

3. The more favorable the terms of trade, the higher the level of em- 
ployment may rise without inflationary pressures. 

4. The more compressible profit margins, the higher wages may be 
pushed without price increases. 

5. The more adaptable or flexible the wage structure, the higher the 
level of employment that may be reached without inflationary develop- 
ments. In a flexible wage structure, wage increases in one sector need not 
result in wage increases in other sectors. 

6. A policy of wage restraint adopted by trade unions or by government 
policy will normally raise the critical level of employment at which wages 
and prices rise. 

7. The response of the labor force to an increase in demand for labor — 
the greater the mobility and the lower the terms on which additional people 
and hours of part-time work can be drawn into the labor market — the 
higher the critical level of employment at which wages and prices rise. 

8. The methods of wage payment, including the prevalence of incentive 
methods of pay, will help to determine whether wages rise early or late 
during an expansion in employment. 

9. The more specific the direction and the impact of fiscal and monetary 
policies, the higher the level of employment that can be achieved without 
inflationary developments. 

Such statements have merit as a starting point for general policy dis- 
cussions since they suggest the diverse fields in which contributions may be 
made by structural adaptations to raise the inflation threshold of the 



18 The Theory of Wage Determination, John T. Dunlop, ed., London, St. Martin's 
Press, 1957, p. 417. 

150 



system. At comparable levels of output, employment, labor peace and 
direct government controls, the inflationary potential of the system may be 
reduced in a variety of ways noted in the above propositions. 

A more constructive policy discussion may begin with an analysis of 
the strategic sectors of the economy which have been particularly prone 
to inflation, or which have not made potential contributions to price 
decreases, or which diffuse wage and price increases widely throughout 
the system. In this view of the structure of the economy, and of the 
mechanics of long-term inflation, all sectors of the economy are not 
equally important. Some are more significant than others in generating 
and diffusing inflation. If secular inflation is to be curbed, without the 
cost of large unused resources, institutional changes are required in such 
strategic sectors. The basic view is that secular inflation is to be tackled in 
policy terms at these strategic sectors of the economy, although not neces- 
sarily by governmental action and certainly not necessarily by legislation. 

Input-output relations 

In seeking to identify such strategic sectors of the economy, the concept 
of input-output tables has been utilized. 19 If one assigns the inputs of 
materials and services into each industry to the sectors from which they 
are derived, and in turn if one assigns the outputs of each industry and 
traces them through the system to their points of use, it is clear that the 
output of some sectors is widely diffused while the output of other sectors 
may flow outside a given sector in only very small quantities and to only 
a few other sectors. Some outputs are more largely directed to households 
than to intermediate stages. Common sense suggests that coal mining, basic 
steel, power generation and construction can be expected to have more 
diffused effects throughout the economic system than the amusement in- 
dustries, education, restaurants or tobacco manufacturing. Common sense 
also suggests that agriculture has more direct effects on households than 
on intermediate stages. 

The input-output tables enable us to show the structural impact of any 
one sector upon the system more generally by computing two measures of 
significance. The first might be called the household effect or its impor- 
tance to the expenditures of households including indirect effects. Thus, 
the significance of steel to households would sum the effects through auto- 
mobiles, household appliances, razor blades, canned foods, toys, etc. The 
combined relative significance of steel in the Consumer Price Index on this 



19 See, W. Duane Evans and Marvin Hoffenberg, "The Interindustry Relations Study 
for 1947," Review of Economics and Statistics, May 1952, pp. 97-142. 

151 



basis is only a few percentage points. 20 The second measure of significance 
would indicate the ratio of intermediate deliveries of a sector to its total 
gross output and might be called the diffusion effect. The significance of 
steel on this basis would be very much greater since the interest is its 
widespread diffusion as a raw material and semi-finished good through- 
out the economy. 21 

A sector which has a high diffusion rating may be expected to spread 
cost and price increases widely through the system unless such increases 
are otherwise offset. The rate of increase in productivity in these strategic 
sectors with a high diffusion rating is vital to the spreading or dampening 
of inflation. 

The fifteen industries which show the highest diffusion rating out of the 
forty-five industry table in order of greatest diffusion are as follows: 22 

Scrap and miscellaneous industries 

Fabricated structural metal products 

Business services 

Nonferrous metals 

Iron and steel 

Lumber and wood products 

Paper and allied products 

Coal, gas and electric power 

Motors and generators 

Stone, clay and glass products 

Metal-working machinery 

Other fabricated metal products 

New construction and maintenance 

Other electrical machinery 

Chemicals 
The five industries which show the lowest measure of diffusion are as 
follows, starting with the very lowest: Medical, educational and non-profit 
organizations; eating and drinking places, amusements, apparel and rental. 
The relative effects of different industries on the Consumer Price Index 
are shown in the following tabulation. The first column shows the fifteen 
industries with the largest total requirements in the goods and services 
which enter into the Consumer Price Index. The second column shows the 



20 Jules Backman, "Importance of Steel Prices in Consumer Price Index," United 
States Steel Corporation, Steel and Inflation, Fact vs. Fiction, New York, 1958, 
pp. 75-81. 

21 Almost 95 per cent of gross output consisted of inter-industry deliveries. 

22 I am indebted to Professor Alfred H. Conrad of Harvard University for the 
computations of the ratio of inter-industry deliveries to total gross output from the 
1947 table. 

152 



fifteen industries with the largest total requirements multiplied by wages 

per dollar of output. 

(1) (2) 

Food and kindred products Agriculture and fisheries 

Agriculture and fisheries Food and kindred products 

Non-distributed Non-distributed 

Personal and repair services Chemicals 

Apparel Other machinery 

Products of petroleum and coal Nonferrous metals 

Rentals Other fabricated metal products 

Chemicals Other electrical machinery 

Motor vehicles Apparel 

Textile mill products Miscellaneous manufacturing 

Coal, gas, electrical power Other transportation 

Eating and drinking places Products of petroleum and coal 

Railroad transportation Iron and steel 

Medical, educ. and nonprofit Coal, gas, electrical power 

Trade Medical, educ. and nonprofit 

Impact of any sector 

The contribution which any sector makes to cost-price movements of the 
system as a whole can be regarded as comprised of the following com- 
ponents: 

(1) The direct effect of a wage-rate change on other wages in the 
system. These are the impacts through the labor market, the collective 
bargaining mechanism, the influence of key bargains and wage leadership. 
Special interest centers on those wage bargains which tend to spread in an 
interdependent wage structure. 

(2) The direct effects of wage-rate increases on total costs. This impact 
of a wage-rate increase depends on two factors — the ratio of labor costs 
to total costs and the rate of increase in labor productivity. When labor 
productivity is increasing rapidly and wage costs are a very small fraction 
of total costs, the cost impact of a wage-rate increase in one sector on the 
system is mitigated. It is recognized, of course, that in some types of 
markets the gains in productivity might have yielded lower prices rather 
than higher wage rates. 

(3) The impact of other cost components on total costs, and, 

(4) The impact of demand for the output of the sector on prices given 
the market characteristics of the products of the sector, that is, the number 
of sellers, availability of substitutes, entry conditions, product standardiza- 
tion and other well-known features of a market structure. 

We should expect the most serious contribution to inflation to arise in 
sectors (a) which diffuse output widely throughout the system, (b) where 

153 



demand is inelastic with respect to price, (c) where labor costs are a 
relatively high proportion of total costs and where productivity is increasing 
much less than the average, and (d) where wage-rate increases have a 
tendency to spread widely throughout the wage structure of the economy. 
Ideally, we should analyze each sector of the economy from the vantage 
point of these tests. 

Only a few of the most strategic sectors are selected here for brief 
comments. But this discussion illustrates the type of approach, when imple- 
mented with a comprehensive inquiry, that is advocated as policy oriented 
toward secular inflation. 

1. Agriculture — It may seem strange to select agriculture as a strategic 
sector for this purpose even though the products of agriculture spread 
moderately throughout the system and constitute a significant proportion of 
the expenditures of households. Prices received by farmers did not even 
increase in the period 1953-58; they actually declined by 1.2 per cent 
compared to an increase of 8.3 per cent in the wholesale price of all com- 
modities; 22.1 per cent in the wholesale price of producer finished goods 
and 7.9 per cent in the Consumer Price Index. But farm output per man- 
hour increased 25.2 per cent in these five years, materially above the rate 
of increase in manufacturing. Moreover, public policies alone were re- 
sponsible for maintaining agricultural prices at their levels. Professor 
Slichter has said: 

Certainly the program of supporting the prices of farm products is one of the 
most ridiculous ventures ever undertaken by any government anywhere — truly 
Alice-in-Wonderland economics. The Federal government is undertaking to 
prevent the great technological revolution in agriculture from depressing the 
relative prices of a few favored crops — and the American people are docilely 
permitting themselves to be taxed to support this madness. No one has ever 
explained why wheat growers, cotton growers, tobacco growers, corn growers, 
or peanut growers are entitled to have the country taxed to buy their output. 
It makes no more sense than would an undertaking by the government to buy 
up other surpluses, such as empty seats in railroad trains, street cars, or moving 
picture houses, or to buy the surplus output of buggies, harness, old-fashioned 
coal furnaces, moustache cups, and a vast variety of goods that the public no 
longer wants except in very limited quantities. 23 

Had different public policies been followed with respect to agricultural 
price supports, it is not unreasonable to estimate that the price of the food 
component in the Consumer Price Index would not have increased at all, or 
might well have declined, reducing the rise in the aggregate index by one 
fourth. The estimate is important only to indicate that the magnitude of 
impact of this sector is substantial. Moreover, the prospects of price move- 
ments in the next five years are also likely to be significantiy influenced 



23 "The American Economy — Current Trends, Problems, and Prospects, January and 
Early February, 1959," Prepared for Ninon Keizai Shimbun of Toyko. 

154 



by agricultural policies. This sector is certainly one not requiring the 
intervention of government for the first time. Rather is the task to secure 
government policies which encourage a redistribution of resources within 
agriculture. A proper agricultural policy will in addition make a significant 
contribution to price stability. 

2. State and Local Government Services — The purchase of goods and 
services by state and local governments increased from $24.9 billions in 
1953 to $39.6 billions in 1958. The national accounts data indicate that, 
in 1958 prices, the cost of these purchases increased from $29.5 billions to 
$39.6 billions, which reveals an increase in an implicit price index of 
18.5 per cent in this five-year period. 24 Such price increases clearly ramify 
widely throughout the system. 

These price increases for local government purchases are derived in part 
from the fact that municipal wages and salaries for many years lagged 
behind those in other sectors; in more recent years the compensation of 
teachers, firemen, policemen, and other employees has been increased more 
substantially. The period of relatively high employment levels, the growth 
in the scale of government operations, the spread of more modern per- 
sonnel practices into municipal and state governments and the expansion 
of union organization into these governmental units have all been factors 
at work. Moreover, in service-type industries it is difficult to calculate the 
productivity offsets, if any, to increases in the prices of services purchased. 
These higher prices of government services have been diffused in this 
period not only by increases in property and income tax rates adding to 
the costs of intermediate producers, but also by the expansion in excise 
taxes to final consumers which directly affect the Consumer Price Index. 

Any view of the future suggests that the prices of services purchased 
by state and municipal governments are likely to continue to rise as wages 
and salaries in this sector adjust to the levels already achieved in other 
private and public sectors and as the competition for skilled technical and 
professional manpower increases generally in the next decade. The trends 
in financing local and state governments increasingly by sales and excise 
taxes suggest further direct impacts upon prices to final producers. Here 
is a sector which is likely to contribute to secular inflation for a long time. 
Improvements in quality of government services do take place, contrary 
to much popular opinion, but there is apparently no simple way to dis- 
tinguish a price increase from a quality increase at a constant price. Some 
of the secular inflation may really be an improvement in quality. 

3. Basic Steel — The most strategic industrial sector for purposes of 
policy considerations of secular inflation has been the basic steel industry. 



24 Survey of Current Business, National Income Number, July, 1958, Table 8, Im- 
plicit Price Deflators for Gross National Product by Major Segments, 1929-57. 

155 



A vast literature and controversy have grown up around recent price and 
wage movements in this industry. The unions blame the price policies of 
the industry. 25 The industry regards itself much more a "victim rather than 
a cause of inflation" 26 and stresses the role of wage-rate increases which 
rose more than the increase in productivity. The academic profession is 
divided, some taking one side or the other and others taking the attitude 
of "a plague on both your houses." We are likely to see this controversy 
become even more intense. 

The essential statistical facts are fairly simple, although the choice of 
periods in which to make comparisons has often been adapted by the 
parties to suit the argument; the following comparisons are made for three 
different periods: 



Percentage Increase 



Average Hourly Earnings, Basic Steel 
Average Hourly Earnings, Automobile 
Average Hourly Earnings, Durable Goods 
Average Hourly Earnings, All Manufacturing 
Consumer Prices 

Wholesale Prices, All Commodities 
Wholesale Prices, Producer Finished Goods 
Wholesale Prices, Steel Mill Products 



1953-1958 1947-1958 


1939-1958 


33.8 100.8 


244.9 


18.7 72.4 


177.6 


21.9 76.5 


226.6 


20.3 72.2 


236.5 


7.9 29.2 


107.7 


8.3 23.7 


137.9 


22.1 62.0 


n.a. 


34.6 108.6 


174.4 


Percentage Increase Per Year 



1953-1957 1947-1959 1939-1957 



Output per man-hour, 

Private domestic economy 27 
Unweighted 2.8 4.0 4.1 

Weighted 2.8 3.5 3.1 

1947-1955 1939-1955 
Output per man-hour 
Steel 2 * 2.8 2.7 

In the period 1953-58 steel wages rose very significantly faster than 
automobiles 29 and the average of all manufacturing and durable goods 
industries. Steel wages were rising materially faster than the rise in pro- 
ductivity in steel or the private domestic economy. Steel prices have risen 



25 Industrial Union Department, AFL-CIO, Labor, Big Business and Inflation, Sep- 
tember, 1958, pp. 18-23. 

26 United States Steel Corporation, Steel and Inflation, Fact vs. Fiction, 1958, p. 91; 
Roger M. Blough, Inflation as a Way of Life, November 9, 1956, p. 12. 

27 Solomon Fabricant, Basic Facts on Productivity Change, New York, National 
Bureau of Economic Research, Occasional Paper 63, 1959, p. 45. 

28 Bureau of Labor Statistics, Man-Hours Per Unit of Output in The Basic Steel 
Industry, 1939-55, Bulletin 1200, Washington, D. C, September, 1956, p. 6. 

29 This experience is the more interesting in light of the steel industry rejection of 
the Wage Stabilization Board's recommendations in 1952 which sought to restore a 
relationship between wage rates of steel and automobiles. 

156 



very sharply more than all wholesale prices and even more than producer 
finished goods which are much influenced by steel. After surveying various 
industrial sectors, Otto Eckstein concluded: "The chief bottleneck appears 
to have been in the steel industry. Here prices and wages rose most." 30 

The same general conclusions are appropriate to a lesser degree for the 
period 1947-58. When one considers the period of the last twenty years as 
a whole, the steel picture is less distinctive, since during the war period 
wages and prices did not move up in steel as much as in other sectors, for 
a variety of reasons. There arises the question whether the period 1953-58 
represents a new and a continuing sector problem for the economy or 
largely the realignment of longer-term relations. 

Basic steel is a pivotal sector from the analysis of the spreading effects 
of inflation because: (1) it rates so high in the diffusion of its output; 
(2) demand for steel is said to be inelastic for most uses; (3) labor costs 
are a significant proportion of total costs, and productivity has not in- 
creased materially faster than the average of industry; and (4) the steel 
wage contract is a key bargain, along with automobiles, in the whole heavy 
goods sector, and steel wage rates exercise influence on other wages in such 
areas as Chicago and Pittsburgh. 

There seems to me little point in getting into the argument between the 
union and the industry as to whom to blame for the experience of 1953-58 
particularly. It is important to understand that the result has followed 
from each operating within its own framework. In the industry's view it 
has sought valiantly to resist unwarranted wage increases by five nation- 
wide strikes since the end of World War II. It has had to push up prices 
in order to meet these higher wage costs, the increasing costs of replace- 
ment of equipment not allowed in depreciation for tax purposes, and to 
secure profits to meet the needs for expansion. The union on the other 
hand has chosen to emphasize the oligopolistic power of the industry to 
set prices and to assert that profits warranted higher wage rates, particularly 
when the industry was operating at peak capacity as it was during the 
period surrounding the decisive last contract negotiations in 1956. I do 
not believe that theoretical understanding or policy wisdom is to be found 
in pursuing the arguments of whether the pot or the kettle is blacker. 
Product market and labor market power of buyers and sellers are highly 
interdependent. 

My analysis of the decisive mechanism of wage and price movements in 
basic steel in recent years may be stated briefly, recognizing that other 
factors may have been operative. The central problem of the industry 



80 "Inflation, The Wage-Price Spiral and Economic Growth," The Relationship of 
Prices to Economic Stability and Growth, Compendium of Papers submitted by 
Panelists Appearing Before the Joint Economic Committee, Washington, D. C , 
1958, p. 370. 

157 



bearing on inflation has been how to finance expensive new capacity in an 
industry where increases in productivity have not been significantly dif- 
ferent from the average of manufacturing. If the industry is to finance 
new capacity substantially from within, 31 prices must be increased to pro- 
vide the profits for expansion. But such profits necessarily stimulate higher 
wage increases than would otherwise take place. This has been the central 
dilemma, and the fundamental reason which led the financial interests in 
the industry in the 1956 negotiations to seek a five-year contract. 

The community has a major problem in the steel industry if it continues 
to demand additional capacity, as is indispensable in the context of both 
the international obligations and economic growth. The expansion of steel 
capacity, financed by the industry, generates wage and price increases in 
this strategic industrial center. From the point of view of short-run public 
policy, it seems to me there is a particularly strong case for the reexamina- 
tion of the issue of financing steel capacity by rapid amortization or by 
other methods used in wartime. From the longer-run point of view, the 
gradual growth of effective substitutes for steel, a process under way in 
a number of industries, is likely to prove the most effective deterrent to 
wage and price increases in this sector. The community will slowly become 
less dependent on steel, and a relatively smaller proportion of its materials 
requirements will be provided by steel. The growth of concrete structures 
in construction and the expansion of aluminum and plastics in manufactur- 
ing are examples. I have little confidence in seeking to affect price and 
wage movements by public policies designed to alter product or labor 
market structures. 

The contrast between basic steel and two other industries which ramify 
widely throughout the system will underscore the combination of circum- 
stances which has characterized steel in this period. In the postwar period 
(1947-57) wages and prices in bituminous coal have both increased — 
wages particularly sharply, although not quite as rapidly as in steel (85 per 
cent compared to 100 per cent). But prices in coal increased only one- 
third the percentage rise in steel, and productivity increased more than 
double that of steel. The substitution of other fuels for coal, the rapid rise 
in output per man-hour and the absence of the problem of expansion in 
heavy cost capacity made the difference, particularly in the latter part of 
the period. The electric power industry is another which ramifies strategic- 
ally throughout the system, but the effect of wage rate increases on prices 
has been mitigated by both the small proportion of operating labor 
costs to total costs and by the rapid growth in size in generating plants 
and productivity. 



31 The argument may be made that steel capacity should have been financed to a 
greater extent by external borrowing, but new public policy in this direction is 
not likely. 

158 



4. Construction — Another strategic sector which diffuses widely 
through the system is the building and construction industry. In the period 
1953-58 construction wages increased 24.6 per cent, and in the period 
1947-58 the increase was 83.8 per cent, almost the same as in bituminous 
coal mining. Construction costs increased but not to the same degree as 
the rise in producer-finished goods. This sector is particularly significant 
for growth, for it also strategically influences the cost of investment goods. 
The industry has a high degree of internal competition, not alone among 
different enterprises in given specialties, but also among different types of 
contractors and different methods of construction and materials. 

The unions and the contractors in this industry are in the process of 
establishing a new joint machinery to provide a continuing forum in which 
all labor-management problems of the industry may be considered (which 
are not provided for by existing machinery). The objective is to promote 
the industry and to work systematically on such questions as difficult local- 
ities, work practices, training, labor productivity, and a host of other issues 
affecting costs and efl&ciency. The response of costs to increased demand 
in this industry is very significantly determined by such practices and 
institutional arrangements. The possibility of making this sector of the 
economy less inflation prone in the long term lies in dealing with such 
detailed problems. 

The inflation potential of the construction industry, as many others, 
cannot well be dealt with by general monetary policy, which cuts volume of 
new construction, but rather by a direct approach to the real problems 
of efficiencies. There are few short cuts to the control of secular inflation. 
It seems to me that slow and hard work in institutional transformations is 
the path likely to lead to most constructive results in the long run. 

Summary 

1. This paper has been concerned with longer-term movements in wages 
and prices in contrast to the problems associated with war and postwar 
periods. The interest has been with the period 1953-58 and projections 
for the future. 

2. The relative preferences of the community for production, employ- 
ment, price stability and labor peace are vital to policy prescriptions. 
These objectives are not entirely consistent. There is no great overriding 
preference discernible in the American community to pay a higher price for 
stability. From my own preferences, the community in recent years has 
undervalued production, capital formation and employment in part because 
it has been concerned with inflation. Our international position desperately 
requires us to push these objectives more strenuously and thereby to make 
the inflation problem the more serious. 

159 



3. Legislation, to revise the internal procedures of labor organizations 
or the scope of bargaining, should stand apart from the inflation problem. 
Such measures, without regard to their merits on other grounds, are not 
likely to result in more stable wage rates and prices. These measures 
simply will not create the results on the inflation problem which is claimed 
for them. 

4. An annual conference of labor and management representatives to 
discuss the economic climate of wage negotiations is suggested each year 
(following the Economic Report of the President) to be convened by the 
Secretary of Labor. The Administration should exchange with representa- 
tives of the parties statistical information and views on the short-term 
and longer-run economic outlook. These sessions should not be concerned 
with any particular contract negotiations; they should provide detailed 
discussion of the economic outlook of the community within which par- 
ticular negotiations take place. Moreover, the proposal conceives the role 
of government to provide leadership in ideas, stimulate discussion and to 
make general proposals rather than simply to legislate or regulate in the 
area of economic policy. 

5. For the purposes of policy formation treating secular inflation, it is 
important to analyze the structural characteristics of various sectors of the 
economy. Input-output tables permit the measurement of the impact of 
any sector upon households through consumers goods, and the diffusion 
effect of a sector is indicated by the proportion of gross output which flows 
to other sectors. 

6. The most serious contribution to inflation arises in those sectors 
which diffuse output widely throughout the system, where demand is in- 
elastic with respect to price, where labor costs are a relatively high pro- 
portion of total costs and where productivity is increasing much less than 
the average, and where wage rate increases have a tendency to spread 
widely throughout the wage structure of the economy. 

7. Secular inflation is to be approached in terms of strategic sectors of 
the economy rather than in terms of further applications of general mone- 
tary or fiscal policy. As an illustration of this method, the problems of 
agriculture, local governments, basic steel and construction were briefly 
discussed. Institutional changes of various forms in such sectors may be 
expected to reduce the inflation potential of the system, or raise the 
threshold to inflation as the system moves at high and increasing levels of 
output. These strategic sectors may be expected to change over time under 
the impetus of different types of expansion; and so continuous and detailed 
reappraisal of the economic structure is required. There are few short cuts 
to control of secular inflation. 



160 



Address to the nation from Arden House: 

Productivity and the consumer 



James P. Mitchell 



Dr. Wriston, I am glad to have this chance to talk with you and this 
impressive group of experts at this Assembly, and at the same time, by 
radio, with a lot of other people who have a big stake in the outcome 
of your discussion. 

This stake is the stake of the consumer, you and me, 170 million of us, 
going to the store for a shirt or a pair of shoes, going to the market for the 
week's groceries, buying another chair for the living room, or, maybe, 
buying a new stove or even a car, paying the dentist's bill, or a tuition fee 
for school. As consumers a lot of us are concerned with prices and our 
ability to pay them with the wages we earn. 

Reading over the learned and impressive papers prepared for this meet- 
ing, I felt that I could contribute most by talking from the consumer's point 
of view. The consumer isn't heard too often, but his actions, in the market 
place, add up to a series of demands which all of us in Government, in 
labor or management, must heed. 



At the Fifteenth Assembly, James P. Mitchell, United States Secretary of Labor, 
delivered this address, on a nationwide radio broadcast. 

161 



First, our economy must continue to grow, and at the same time we 
must have a reasonably stable level of prices. 

Second, we must increase our productivity, to achieve that growth, to 
permit that price stability. 

Third, labor and management must bargain together responsibly and 
reasonably, so that our products can compete in present markets, and for 
new markets. 

These must become new American imperatives. Most of the postwar 
period until recently has been described as a seller's market. It has been a 
time when the American producer had a very receptive market both at 
home and abroad. 

Shortages of goods during the war had helped to build up a great backlog 
of demand. Washers, refrigerators, radios, and many other appliances were 
hard to get during the war. Automobiles were not being made. Housing 
was scarce. People had saved their money during wartime and looked 
forward to buying. The postwar surge of pent-up demand lasted a long 
time. And another surge of buying occurred during the Korean War. The 
American consumer was easy to please. But recently the consumer is look- 
ing more carefully before he buys, and grumbling about prices, and this is 
all to the good. 

Internationally, the world markets which American producers could once 
fill with our products, at our prices, are no longer reserved to us. Plants in 
Germany, Japan, Britain, France, Italy which were bombed out or didn't 
exist in 1945 are now producing products to compete with ours. Theirs is 
a stiff challenge. Look at the brands of typewriters, dishes, woolens, and 
automobiles you see in our shop windows and on the streets. On the other 
hand, some of our products are in very great demand abroad: we are 
still selling more overseas than we are buying. But we are going to have 
to sell harder, against tougher competition, to stay in these markets. 

American workers and American industry have a big stake in our ability 
to sell products in foreign markets. This ability depends upon our keeping 
costs and prices down. Upon this ability will depend not just present jobs 
but new jobs we will need in the future. 

At home consumers will buy selectively, looking for quality, and they 
are increasingly conscious of price. 

Since we are talking about the consumer's interest, we should be more 
conscious of what prices have done, here, in our own home market. Since 
World War II, prices have been following a staircase upward. They have 
leveled off during recent periods, but taken together they have not gone down. 

Under these circumstances, some buyers in some markets have shifted 
to other products or have not bought at all. We have, for example, lost a 
part of the domestic market for automobiles. Some people could not find 

162 



a car they wanted at a price they could pay, so they either didn't buy a car 
or bought a foreign car. 

When the consumers don't buy because prices have gone too high, pro- 
ducers don't produce as much, and that means they don't employ as many 
people. Higher profits or higher wages, resulting in higher costs and prices 
that consumers won't pay, mean that some people may pay with their jobs. 
Workers and management have to recognize that consumers may not be 
willing to follow prices upward indefinitely. 

Some people whose incomes have not gone up have been feeling the 
pinch of high prices. People planning for retirement, people saving to 
educate their children will feel that pinch even more later. They want and 
need a stable-price level. We all want and need growth. The question is, 
how can we get both. 

Two approaches are usually suggested. The first is greater production 
and improved productivity. This is, I believe, a sound approach. A second 
idea is heard in some places : it calls for Government to move in on wages 
and prices. This approach is not, I believe, a wise one, for reasons I want 
to discuss in a moment. But first, let's look at productivity. 

Productivity, in its simplest terms, means the amount of a product a 
worker turns out in an hour — how many bricks, sheets of steel, or shirts, 
how many columns of figures added. 

More productivity means turning out more product in the same working 
time. Usually this involves better machines, better-trained workers, and 
improved management. Productivity improvements in factories, mines and 
on farms have made possible much of our great national prosperity in the 
past. Out of increased productivity must come further increases in our 
standard of living. 

Increased costs should be covered by increased productivity, for, if they 
are not, prices rise. Some people think that the benefits of productivity 
increases should be divided only between labor and management, but there 
is a third party, the consumer. The consumer should share in increased 
productivity through better quality goods or lower prices. This is why 
increases in labor costs for the economy as a whole ought to be so related 
to productivity improvement that increases in price levels will not result. 

Applying this general principle to specific bargaining situations is diffi- 
cult, and I want to say just a word about it. National productivity figures 
are averages drawn from many different producers: farms, factories, mines, 
offices. Applying such averages mechanically is like taking the average 
size of a dozen assorted men and making them all wear the average size 
suit. The same is true even of industry-wide figures : as labor and manage- 
ment well know, there are great differences in the economic situation of 
different companies and plants within any one industry. 

163 



In arriving at what might be a proper wage increase in a particular in- 
dustry or enterprise, management and union negotiators have no simple, 
ready-made formula. They should be extremely cautious in the way they 
use productivity figures. There are many factors other than productivity 
which determine whether a wage settlement is fair in a specific industry. 
There is also a national interest at stake. 

Certainly labor and management in the major industries must realize 
that their wage and price actions are often followed by other industries 
whose situations may be vastly different, with the result that wages and 
prices go up all along the line. This has frequently been the case. 

Labor and management have a tough assignment in arriving at fair wage 
settlements which will not cause price increases. Workers and union leaders 
must be more concerned about real wages, not merely money wages. When 
making wage demands they should consider what these wages can buy. 

Another challenge that lies ahead for labor leaders is to be more constant 
in conveying to their members what rising costs and prices and the goods 
they produce can cost them. Wage increases that result in price increases 
can cut the market for the goods they produce, and, ultimately, cost jobs. 

The challenge that lies ahead for management is just as difficult. It is 
not always possible to take the easy way out of passing along all cost 
increases to the customer. Some must actively seek ways to trim inefficient 
practices and tighten up to face competition. Others must recognize that 
lower and more flexible prices can improve sales and create jobs. It may 
even mean reassessing attitudes toward profit margins. It will certainly 
mean talking frankly about the effect of wage-cost increases, discussing 
fully the economic facts with union representatives. 

This brings me to the last point I want to make: Labor and manage- 
ment should not depend upon Government action — involvement in col- 
lective bargaining — to relieve them of any of their responsibility to bargain 
reasonably. 

This is true for two reasons; first, experience shows that when the Gov- 
ernment does step in it keeps on walking into a much wider field than just 
a wage settlement; second, experience also shows that when a third party 
is making the final decisions, the other two parties whose interests are at 
stake act differently than they would were they making the decision them- 
selves. They get reckless. They exaggerate their claims and charges. After 
all, what have they got to lose? 

Ultimately if enough decisions are taken out of the hands of both the 
workers and the employers, the Government assumes the responsibility for 
finding just solutions. What may follow is that Government finds itself 
setting wage scales, fixing prices, profits and the conditions of work, de- 
termining hours, hearing grievances, and throwing its weight around in 
other private matters. 

164 



This should be especially significant to a society like ours that is rooted 
in free institutions. We have to consider and preserve the freedom of 
negotiation and agreement. This freedom depends upon its responsible 
exercise. Labor and management have an obligation to work toward settle- 
ments fairly and objectively, in full possession of the facts. They must 
recognize that they are not alone at the bargaining table: the customer, 
the public, are also there. 

The peaceful settlement of negotiations is one measure of labor-manage- 
ment progress, but is meaningful only if achieved with full recognition of 
the public interest in the growth of our economy and in stable prices. 
Labor and management, especially in the industries at the heart of our 
economy, are, I think, becoming more responsible and realizing that then- 
actions have consequences far outside their own pay envelopes and profit 
margins. 

Among these consequences are those I have been discussing tonight. 
Consumers, we consumers, are getting very suspicious of further price in- 
creases, and we believe we have the right to demand a greater share in the 
benefits of increased productivity. And, as workers, we recognize two 
things — first, that jobs may be lost in some industries if prices get so high 
that consumers may not buy the goods they produce, and, second, that 
rising prices shrink the value of our own paychecks. 



165 



Address to Fifteenth American Assembly: 

Labor costs and prices 



Sumner H. Slichter 



I start with the elementary proposition, too often overlooked, that the 
principal purpose of an economy is to produce and distribute goods. Its 
purpose is not to produce a stable price level — though it does a better job 
of distributing goods when the price level is stable. But keeping the price 
level stable may interfere with producing the maximum quantity of goods. 
Hence, there may be a conflict between the best possible job of production 
and the best possible job of distribution. 

The economy has developed the kind of institutions that tend to produce 
a slow rise in the price level. One of these is technological research which 
stimulates demand partly by developing new investment opportunities and 
partly by developing new consumer goods thereby keeping down the rate 
of saving. Another price-raising institution is the trade union. Even before 
the days of trade unions, there was a strong tendency for technological 



Sumner H. Slichter, Professor of Economics at Harvard, has a long and dis- 
tinguished record in education and public affairs as scholar, author, teacher, con- 
sultant, and director. 

167 



change to produce higher money wages rather than lower prices, but the 
trade union greatly increases the tendency of wages to rise both in response 
to increases in demand, and even in the absence of increases in demand. 1 

Although the economy has developed a strong tendency to produce a 
slowly rising price level, the problem of inflation has turned out to be less 
serious than important government officials and businessmen would have 
the country believe. Ill-informed talk about inflation has aroused unjusti- 
fied fears and has given Americans an inferiority complex on the subject 
of inflation that the facts do not warrant. Few Americans are aware that 
the rate of price increase is diminishing. The Consumer Price Index rose 
73 per cent in this country in the decade 1939 to 1948 and only 20 per 
cent in the period 1948 to 1958. In the five years ending in 1953 the 
Consumer Price Index rose by 11.3 per cent; in the five years ending in 
1958, by only 7.8 per cent. Few Americans are aware that the country has 
done a far better job of limiting the rise in the price level than have most 
other industrial countries. Few Americans know that the index of con- 
sumer prices has risen in the United States during the last ten years less 
than one-third as much as in the United Kingdom, only one-fourth as 
much as in France, one-seventh as much as in Austria, one-third as much 
as in Norway, less than half as much as in Sweden, the Netherlands, or 
Denmark, and over one-third less than in Italy or Canada. 

The predictions that creeping inflation would produce dire consequences 
have turned out to be untrue, or the predicted results have turned out to be 
avoidable or not serious. For example, the country has been told again 
and again that the expectation of inflation would produce an accelerated 
rise in prices that would ultimately end in a crash; that inflation would 
cause the United States to be priced out of world markets; that inflation 
will impair the quality of business decisions; that inflation will seriously 
injure recipients of fixed-dollar incomes and owners of fixed-dollar assets. 

Events have shown the error in the theory that creeping inflation will 
soon break into a gallop. Among 16 industrial countries in Europe and 
North America the rise in the Consumer Price Index between 1953 and 
1957 was less than the rise between 1948 and 1953 in all cases except two. 
The two exceptions were Belgium and Switzerland, where the rise in the 
Consumer Price Index in both periods was small. In the United States 
the rise in the consumer price level in the second period was less than half 
the rise in the first period. Even in countries such as Austria, France, 



1 Many supply prices, especially the supply price of labor, are sensitive to shifts in 
demand. Let the demand curve for labor shift to the right and the supply curve will 
shift to the left — not because the unemployed hold out for more, but because the 
employed demand more. It is a mistake to assume that the unemployed have much 
effect upon wage rates. These rates are influenced by the employed who may or 
may not be influenced by fear of becoming unemployed. 

168 



Norway, Sweden, and Great Britain, where the rise between 1948 and 
1953 was very rapid, the rate of increase did not accelerate — it dropped 
substantially in the subsequent five years. 

Rise in the Consumer Price Index in Sixteen Industrial Countries 
in Europe and North America in Two Recent Periods 

1948-53 1953-57 



Austria 


100.0 


12.0 


Belgium 


5.3 


7.0 


Canada 


19.0 


6.0 


Denmark 


23.5 


16.0 


Finland 


56.3 


20.0 


France 


66.7 


6.0 


Great Britain 


29.9 


16.0 


Ireland 


26.6 


12.0 


Italy 


16.3 


10.0 


Netherlands 


28.2 


8.0 


Norway 


35.1 


12.0 


Spain 


26.6 


23.0 


Sweden 


29.9 


13.0 


Switzerland 


4.2 


5.0 


United States 


11.1 


5.0 


West Germany 


7.5 


6.0 



The most important reason for the rate of increases in prices to drop 
is the simple and obvious fact that it takes money to buy goods. If people 
wish to buy in anticipation of higher prices, either they must draw on 
their idle cash balances or they must draw on their credit. Their ability 
to draw on their cash balances is limited and their ability to draw on their 
credit depends upon the reserve position of the banks. Consequently, 
whether or not creeping inflation becomes a gallop rests in the last analysis 
with the monetary authorities. Unless they see fit to provide the necessary 
reserves to finance galloping inflation, inflation will never become a gallop. 

Creeping inflation is not likely to cause the United States to be priced 
out of world markets. The superiority of the United States is great, par- 
ticularly in manufacturing, as is shown by the fact that in 1958 our exports 
of finished manufactured goods were 2.4 times the value of our imports 
of finished manufactures. 

No one can be sure what the future will bring, but, as I have pointed 
out, prices in most other industrial countries have risen faster than in the 
United States. Unfortunately, some foreign countries are hampered in 
competing with the United States by various special circumstances. In 
Britain, the largest manufacturer outside the United States in the free 
world, taxes are exceedingly burdensome and efficiency in much of industry 
is held down by wasteful union rules and, in the metal trades, by an unruly 

169 



shop stewards' movement which in many plants has prevented management 
from exercising proper control of operations. 

But American business must expect foreign competition in some fields 
to become stiffer — particularly in products requiring considerable hand- 
work. The stiffer competition will be good for the United States — a useful 
spur to efficiency, a stimulus to the development of much-needed plans for 
training craftsmen, a modest deterrent to increases in prices. The need for 
stronger foreign competition in American markets is shown by the howls 
of anguish which arise whenever a few imports enter the country. Our 
exports of iron and steel products in 1958 were more than twice as large 
as our imports; our exports of machinery were seven times as large as our 
imports; our exports of cotton manufactures were nearly twice as large 
as our imports. Nevertheless, all of these industries have expressed alarm 
over the influx of imports — as if they had the right to sell to American 
consumers unchecked by foreign competition. 

Creeping inflation is not likely to cause a flight from the dollar. The 
gold losses of the last year show in the main that the international monetary 
system is working as it should work and that the United States is playing 
its proper role of banker to the world. 

The people who predict a general withdrawal of balances from the 
United States neglect to say where the money is to go or what the money 
will be looking for. Short-term funds can earn a higher rate of return in 
money markets other than New York, but the great trade of the United 
States creates a demand for dollar balances. New York also offers the 
attraction of a free and stable rate of exchange. If the owners of funds are 
looking for attractive long-term investment opportunities, they can find 
such opportunities in limited quantities in Canada and some parts of 
Western Europe, but the United States offers a combination of good return 
and political security that has strong appeal to conservative investors. In- 
vestors who put their funds in the Near East and parts of Africa and the 
Far East had better plan to recover their original investment rather prompt- 
ly because there is real danger that the investments will be confiscated. 

There is no evidence that creeping inflation or the expectation of it has 
undermined the quality of business decision making, that managements 
have been careless in making decisions on the ground that the rise in prices 
would bail them out. The outcome of competition depends upon conditions 
in specific markets, not on the general course of the price level. Hence, 
business enterprises have had to study specific market conditions as much 
as ever before. There have been important cases of overinvestment (petro 
chemicals and aluminum, for example,) but these cases are attributable to 
the desire of various producers to get a strong position in markets that 
were being made attractive by technological change. The greatest mistake 
in business decision making in recent years was the persistent efforts of the 

170 



Big Three automobile manufacturers to sell longer, lower, and wider cars 
after the public had begun to revolt against the acrobatics required to get 
in and out of these cars, the inconvenience of parking them, and the high 
cost of operating them. But the blunder of the Big Three was not due to 
creeping inflation; it was due to the difficulty of changing the trend of 
thinking in big organizations. 

Creeping inflation tends to diminish the purchasing power of pensions 
of the conventional type. Fortunately, this problem can be dealt with in 
various ways. One way is to base the pension on earnings during the last 
ten years of service. These earnings are usually higher than the employee's 
lifetime average and tend to rise with the price level. Another method is to 
liberalize the pension plan from time to time as hundreds of companies 
have done and as has been done with the federal old age and survivors' 
plan. In the latter plan the average monthly old-age pension was increased 
168 per cent between 1946 and July, 1958. In the same period the Con- 
sumer Price Index rose less than 49 per cent. Finally, the problem may be 
met. in large measure by the use of variable annuities. 

Creeping inflation diminishes the purchasing power of savings accounts 
and life insurance. In a world of perfect markets in which people bought 
and sold with foresight, expectations concerning increases in the price level 
would lead to increases in the rate of interest. A compensating rise in 
interest rates has not yet occurred, but in the course of time it will probably 
occur. Certainly the government officials have been far from frank with 
the people. While attempting to scare them with unrealistic predictions 
about the consequences of inflation, government officials have neglected to 
urge people to stand out for a decent rate of return. A proper rate of 
interest will compensate owners of savings accounts and life insurance 
policies for the decreased purchasing power of their investments. 



II 

The causes of the rise in prices in the last ten years are numerous, and 
I do not intend to discuss all of them. There was a rise of $324 billion, 
or nearly 75 per cent, in the total net public and private debt between 1948 
and 1958. And yet in spite of this enormous growth in debt, the economy 
was less liquid in 1958 than in 1948. Between 1948 and 1958, the gross 
national product in current dollars increased 69 per cent; in dollars of 
constant purchasing power, 35 per cent; and the money supply (demand 
deposits and currency outside of banks) 28 per cent. The annual rate of 
turnover of demand deposits in 337 or 338 reporting centers increased 38 
per cent between 1948 and 1958. The rise in the rate of turnover shows 
that there has been considerable inflation originating in demand even in a 
period when rising costs were pushing prices up. 



171 



The aspect of inflation that I wish to discuss is the influence of trade 
unions. This influence is broader and considerably different from the role 
usually attributed to trade unions. A realistic view of the influence of 
unions leads to changes in conceptions of the economic theory of unions and 
important modifications in the theory of wages and the theory of employ- 
ment. Unions are not only a major reason why labor costs have risen in 
the last ten years about twice as fast as output per man-hour, but they are 
important generators of income thereby tending to raise the demand for 
goods as well as the cost of goods. 

The traditional view of economic theory has been that the success of a 
group of employees in enforcing a higher supply price for their labor in the 
absence of changes in the demand for their services simply means a re- 
distribution of incomes to the advantage of those members of the wage- 
increasing group who succeed in keeping their jobs. What these persons 
gain, others lose either in the form of less employment, lower profits, or 
lower wages. 

But the traditional analysis is incomplete — it overlooks the fact that 
bargained wage increases which occur in the absence of increases in 
demand, frequently, though not always, raise the total volume of spending 
in the economy sufficiently to maintain or even increase the total volume 
of production and employment. Hence, bargained wage increases do not 
merely transfer and redistribute money income; they often generate gains 
in money incomes and production. Three types of cases should be 
distinguished: 

Type One — selling prices are raised as a result of wage increases, 
and the demand for the product is elastic. 

Type Two — selling prices are raised as a result of wage increases, 
and the demand for the product is inelastic. 

Type Three — selling prices are not changed as a result of the wage 
. increases. 

Type One — selling prices are raised by the firm, but demand for the 
commodity is elastic. In this case the effects of the wage increase are 
deflationary. Expenditures by the firm's customers for its products drop, 
and some of the money not spent for the products of the firm goes into 
liquid reserves of the customers. The payroll of the wage-increasing firms 
drops too, except in a few freak cases in which the demand for labor in 
the short run is quite independent of changes both of output and of wage 
rates. As a general rule, the drop in employment associated with a drop 
in sales would produce some drop in payrolls despite the wage increase, 
and, in addition, the drop in the total income of the firm would produce 
some drop in the non-payroll expenditures of the firm. Thus the effect of 
wage increases when the demand for the product of the firm is elastic 
is deflationary. 

172 



Type Two — selling prices are raised by the firm but demand for the 
product is inelastic. Under these conditions the wage increase is inflation- 
ary. It leads to offsetting increases in prices, and since the demand for the 
product is inelastic, the price increase will raise the total amount spent for 
goods of all kinds — the product of the firm raising wages and the products 
of other firms as well — since the higher price will ordinarily cause some 
shift in the use of money from speculative uses to transaction uses. The 
increase in spending resulting from higher prices charged the wage-increas- 
ing firm will not, of course, be sufficient to maintain the previous physical 
volume of production. Observation of this fact frequently leads economic 
theorists to reach erroneous conclusions concerning the total effect of wage 
increases. The theorists fail to take account of the expenditures by the firm 
itself (including its employees) as well as by the firm's customers. 

The demand for labor in the short run is almost invariably inelastic. 
Hence, the wage increases will raise the firm's payrolls. Its non-payroll 
expenditures will shrink. Ordinarily, however, the shrinkage in the non- 
payroll expenditures will be less than the expansion of payroll expenditures. 
The enterprise must be expected to use its resources so that for every use 
the ratio of marginal cost to marginal advantage is the same as the ratio 
for every other use. Hence, when outside influences (the union) force the 
firm to increase its payrolls, the enterprise will meet the cost, not solely by 
cutting other expenditures, but partly by drawing on liquid resources and 
partly by greater use of credit. Only in a few extreme cases, where wage 
increases force the firm out of important markets or impair its credit, will 
payroll increases reduce the total expenditures of the firm. 

The increase in the outlays of the wage-increasing firm has the same 
effect on the rest of the economy as any autonomous increase in spending. 
The increase is financed by a draft on liquid resources — a shift of money 
from inactive (speculative) uses to active (transaction uses) or greater use 
of bank credit for working capital. The effect of the autonomous increase 
in spending falls into two parts — the effect on consumption and the effect 
on investment spending by non-wage increasing firms. The effect on con- 
sumption is determined by the marginal propensity to consume in accord- 
ance with the familiar Keynesian multiplier. The effect upon investment 
depends upon the shift in the investment function in the rest of the 
economy. This function is the result partly of the state of liquidity of 
business concerns and partly of the appraisals of the business outlook that 
are constantly being made. 

The combined increase in spending by customers of the wage-increasing 
firm and by the wage-increasing firm and its employees may be expected to 
increase the total amount of spending in the economy more than sufficiently 
to sustain or increase production at the new higher price level. It cannot 
be asserted that this is an inevitable or necessary result, but it can be 

173 



asserted that it is almost inevitable in those cases in which the demand 
for the production of the wage-increasing firm is inelastic. The autonomous 
increase in spending would have to be very small or the multiplier very 
small for the total volume of spending to rise too little to sustain an in- 
crease in physical production. 

Type Three — the selling price is not raised by the firm. If one or a 
minority of several competitors is organized, the firm may find itself com- 
pelled to grant a wage increase that its rivals are not granting. Thus the 
management must choose between raising a price with the prospect that 
the demand will be found highly elastic because rival firms do not raise 
their prices, or of holding the line on prices in spite of the wage increase. 
If the first course is selected, the case becomes one of Type One which has 
already been discussed. If the second course is selected, the situation be- 
comes one of the Case Three type. 

In this third type of case, expenditures of the firm's customers are not 
changed, but there is usually some increase in the outlays of the firm and 
its employees. The demand for labor in nearly all cases is inelastic. Hence, 
payroll expenditures rise. Non-payroll expenditures drop, but not suffi- 
ciently to offset the rise in payroll expenditures. The reason is that cuts 
in non-payroll expenditures can be made only by accepting increasing 
disadvantages. Hence, the enterprise has an incentive to take various steps 
to avoid cuts in its non-payroll expenditures. These steps may include 
drawing on the firm's liquid resources or relying to a greater extent upon 
bank loans for working capital. Thus, there is an increase in spending 
similar to the increase in situations of Case Two. Both consumption and 
investment throughout the economy are stimulated. But the effect on 
expenditures is less than in situations of Case Two. 



Ill 

Whether or not trade unions on balance are instruments of deflation or 
instruments of inflation depends upon the relative importance of the several 
types of case. 

Cases of Type One in which the demand for the product of the firm is 
elastic are found when the firm is exposed to special cost influences that do 
not affect rival firms. An example might be a firm compelled to bargain 
with a union under conditions that prevent the wage settlements made with 
the union to have much effect upon the wages paid by rival firms. Merely 
to describe the situation shows how unusual it is. Most firms are exposed 
to pretty much the same cost influences as their rivals. Hence, all are 
more or less affected alike by changes in costs. This means that all rivals 
make more or less the same adjustments of prices to changes in costs. If 



174 



that is so, the elasticity that counts is the elasticity of demand for the 
product of the industry rather than the elasticity of demand for the brands 
of the several enterprises. The elasticity of demand for the product 
is much less than the elasticity of demand for the several brands and is 
much more likely to be less than minus one. Hence, one concludes that 
cases of Type One are not particularly frequent — at least after unions be- 
come strong and pervasive. There are, however, a few industries in which 
new firms are so easily started that the elasticity of the demand for the 
product, as distinguished from the several brands, is high. Shoes and men's 
and women's garments are examples. 

It follows that cases of Types Two and Three — the situations in which 
unions are generators of income — predominate. I do not think that 
Type Three is particularly numerous. The most common is the Type 
Two situation in which all competitors negotiate wage settlements more 
or less simultaneously and make price adjustments more or less in unison. 
In these situations expenditures for the product are governed by the 
elasticity of demand for the output of the industry rather than by the 
elasticity of demand for the output of individual firms. The elasticity is 
likely to be less than minus one and the wage increases are inflationary. 2 
Examples of Type Two are steel, rubber, automobiles, meat packing, paper, 
oil, farm equipment, much of construction, airplane manufacture, air 
transportation, trucking. 

The conclusion is that trade unions as a rule do more than transfer 
income from some parts of the economy to others. They affect the size 
of the total flows of income as well as the relative size of its components. 
A few of the settlements negotiated by unions are deflationary — they re- 
duce the size of the total income flows. More often the effect is inflation- 
ary — the effect is to increase the size of total income flows. As trade 
unions become stronger and as they become more pervasive, the greater 
becomes the tendency of their wage settlements to affect the prices charged 
by all firms in the industry. Consequently, an important difference between 
a large, well-established and strong trade union movement and a weak, 
poorly established movement is that the strong, well-established movement 
can bargain on the basis of the industry elasticity of demand, whereas the 
weak movement must bargain on the basis of the firm elasticity of demand. 

All of this is a way of saying that as the trade union movement gains 
strength, its economic significance changes. A weak movement may pro- 
duce only a small preponderance of deflationary effects because a large 
proportion of its bargains will be of Type One. The strong trade union 
movement will be inflationary because most of its bargains will be of Type 



2 The amount of investment in most industries is such that the inelastic segment of 
the demand curve extends a considerable distance above the point at which goods 
are priced. 

175 



Two. At present the trade union movement in the United States is suf- 
ficiently extensive and powerful so that most of its bargains are of 
Type Two. Our trade union movement has become a powerful income 
generating instrument — a built-in source of demand for goods and a 
source of inflation. 



IV 

If trade unions are in most instances income generating organizations, 
the economy is stronger than we have supposed it to be. The influences 
making for expansion are stronger than we had supposed them to be. 
Likewise, the influences tending to sustain personal incomes and personal 
consumption expenditures in times of recession are stronger than we had 
realized. What evidence is there that trade unions on balance have become 
generators of income? 

One bit of evidence is the behavior of wages in the face of stationary 
or slightly declining corporate profit rates during the last ten or eleven 
years. The profits of all non-financial corporations as a percentage of sales 
have fallen substantially during the ten-year period. Even the sum of 
profits plus depreciation allowances as a percentage of sales have remained 
about the same with a slight tendency to fall. 3 With profit margins nar- 

Changes in Compensation of Employees Compared with 

Changes in Real Output per Man-hour and with 

Changes in Prices from 1947 to 1958 

Increases in 

average hrly Change in Changes in 

compensation real product Change in Change in wholesale 

of workers of private consumer non-farm prices of 

in private industry per price wholesale finished 

industry man-hour index prices goods 

1947-48 8.5 per cent 3.6 per cent 7.6 per cent 8.5 per cent 7.9 per cent 

1948-49 2.7 2.9 -0.9 -2.0 -2.8 

1949-50 5.7 7.1 0.9 3.7 1.8 

1950-51 9.3 2.5 8.0 10.4 9.5 

1951-52 5.8 2.2 2.3 -2.3 -0.5 

1952-53 5.9 4.1 0.8 0.7 -1.0 

1953-54 3.5 1.8 0.3 0.4 0.3 

1954-55 2.9 4.4 -0.3 2.2 0.2 

1955-56 6.0 0.6 1.5 4.4 2.8 

1956-57 6.0 2.7 3.4 2.8 3.6 

1957-58 3.0 1.0 2.7 0.3 2.3 






3 In the five years ending 1952, the sum of the profits of all non-financial corpora- 
tions plus depreciation after taxes was 5.83 per cent of sales; in the five years 
ending 1957 it was 5.58 per cent. Profits of all non-financial corporations after 
taxes were 4.0 per cent in the five years ending 1952, and 3.2 per cent in the five 
years ending 1957. 

176 



rowing one would not expect wages to be bid up faster than the rise in 
output per man-hour, and yet in the ten years from 1948 to 1958 hourly 
compensation of employees exceeded the gain in real product per man- 
hour in all of private industry in eight years, and for the entire period the 
rise in compensation per man-hour was nearly twice as large as the gain 
in real output — 63.3 per cent against a gain of 33.3 per cent in real output 
per man-hour. A second bit of evidence is the fact that in every one of 
the last 10 years without exception average hourly compensation of all 
employees in private industry rose more than the Consumer Price Index, 
and in 9 out of the last 10 years hourly earnings of all workers in private 
industry rose more than the wholesale prices of finished goods and more 
than non-farm wholesale prices. A third bit of evidence is the tendency 
of wages to continue rising in the face of falling demand for labor, as 
happened in 1949, 1954, and 1958. 4 A fourth bit of evidence is the success 
of unions in pushing up wages in various industries regardless of market 
conditions — the success of the coal miners in enforcing large wage increases 
in the face of falling employment; the success of the steelworkers in raising 
wages in southern can factories far above the market; the success of the 
steelworkers in enforcing considerably higher scales in paper container 
plants of the can companies than are enforced by the paper makers' 
unions in the same companies; the success of the hosiery workers in driving 
virtually all union full-fashioned mills out of business. 



The success of unions in raising wages far faster than the increase in 
productivity has created a difficult problem of explanation for trade unions. 
The unions like to claim credit with their members for raising wages and 
to deny blame with the public for contributing to inflation. Union spokes- 
men argue that prices have risen for reasons independent of wage increases 
and that unions have simply made offsetting increases in wages. Union 
spokesmen argue that wage increases in conjunction with gains in pro- 
ductivity have raised labor costs only about the amount of price increases. 



4 Between 1948 and 1949, employment dropped from 59,117,000 to 58,423,000 and 
the unemployment rate rose from 3.8 per cent to 5.9 per cent. The consumer price 
level dropped by %o of one per cent, but hourly compensation rose by 2.7 per cent. 
Another drop in employment occurred between 1953 and 1954 when there was a 
decrease from 61,945,000 to 60,890,000, and a rise in the unemployment rate from 
4.9 per cent to 5.6 per cent. Again the consumer price level dropped, this time by 
%o of one per cent, but hourly compensation of employees rose by 2.9 per cent. 

A third drop in employment appeared between 1957 and 1958. The decrease was 
from 65,011,000 to 63,966,000, and the unemployment rate rose from 4.3 per cent 
to 6.8 per cent. Despite this substantial rate of unemployment, hourly compensation 
of all workers in private industry rose by 3 per cent, and the consumer price level 
by 2.7 per cent. 

177 



Hence, wage increases have been the result of price increases, not their 
cause. This theory meets certain difficulties. The rise in value product per 
man-hour between 1948 and 1958 is almost exactly the same as the rise 
in labor costs per unit of product. Does this fact mean that unions knew 
the coming change in output per man-hour and the coming changes in 
prices? Otherwise how would the union negotiators know how much of a 
wage increase to bargain for? Since output per man-hour rose by various 
amounts (as little as % of one per cent in 1955-56 to 7.1 per cent in 
1949-50) and since the year-to-year change in the Consumer Price Index 
varied widely (from minus % of one per cent in 1948-49 to plus 8.0 per 
cent in 1950-51), there is no reason to believe that unions can predict 
these changes. 

A simpler explanation that involves no heroic assumptions about ability 
of unions or anyone else to predict increases in output per man-hour or 
changes in price attributes the rise in prices to the rise in labor costs, and 
the rough correspondence between changes in labor costs and changes in 
the price level to the fact that in a consolidated income statement of the 
American economy compensation of employees represents two-thirds of all 
costs — in other words, is twice as important as all other costs combined. 
When labor costs are twice as important as all other costs combined, a 
tendency for labor costs to rise about twice as fast as output per man-hour 
is bound to lead to higher prices. 



VI 

What should be done about the tendency of unions to generate incomes? 
This is not the sole cause for inflation, though in the last few years it has 
probably been the most important single cause. It has been a useful in- 
fluence in important respects — especially in contributing substantially to 
sustaining incomes during periods of recession. The tendency of unions to 
generate incomes is also useful in accelerating recovery in times like the 
present. Finally, the income-generating capacity of trade unions tends to 
stimulate the growth of the economy by accentuating the tendency for 
demand to outrun productive capacity. 

In spite of these important contributions of trade unions to economic 
welfare, an effort should be made to limit wage increases as a general 
rule to increases in output per man-hour. Relying upon wage increases 
to produce autonomous increases in spending creates too many special 
gains for groups in strong bargaining positions. It is better for the economy 
to get its autonomous increases in spending in ways that benefit all groups 
— through tax cuts or planned budget deficits. 

There is no known and proven way of limiting the generation of income 
by trade unions sufficiently to prevent them from raising the price level. 

178 



It has been suggested that unions be deprived of some of their present 
extraordinary privileges, such as their use of coercive picketing or the 
conscription of neutrals in labor disputes. These changes in the law are 
overdue, but they would have little effect upon the outcome of most 
bargains. It has been suggested that the unions be broken up so that there 
would be several in each industry. Unions would lose some of their present 
ability to support strikes by some members while other members work and 
pay dues and special assessments. But there would be rivalries among the 
new unions, and each would feel a strong urge to make a better settlement 
than any of the others. Hence, there is little reason to expect that breaking 
up unions would as a general rule diminish their upward pressure on wages. 

A series of somewhat unrelated steps might add up to a significant re- 
straint on upward pressure on wages. A great expansion in the use of in- 
dustrial engineers producing capital-saving inventions would be useful. A 
larger proportion of capital-saving inventions (the typical product of the 
industrial engineer) would weaken the tendency of technological change to 
increase the demand for labor. Labor-saving inventions are inflationary 
because they increase the demand for labor, and labor-saving inventions 
will probably continue to predominate. Nevertheless their preponderance 
can be reduced. 

We have just been through a period of strong rivalry among trade 
unions. Nearly all the workers who present attractive organizing oppor- 
tunities have been organized. In the next two or three decades rivalries 
will probably subside somewhat, reducing the upward pressure of unions 
on wages. The aims of unions can be broadened and made more construc- 
tive by the adoption of the Scanlon Plan or variants of it — it marks in my 
judgment an important step forward in the art of management and it in- 
creases the influence of trade unions for good. I like John Dunlop's sug- 
gestion of an annual stocktaking of the economic outlook by representatives 
of labor and management in a government-sponsored conference. Unions 
are not sensitive to public opinion, but they are not immune to the climate 
of opinion. Several years might be required before the stocktaking meetings 
become matter-of-fact. As industry and labor acquire economists whose 
loyalties are first of all to economics and only secondarily to business 
and labor, it will be possible to have more or less objective appraisals 
of the economic outlook. But real professionals, who value the good 
opinion of their colleagues more than the good opinion of their bosses, 
will be necessary. 

Finally, in the event that the country becomes seriously interested in 
halting the slow rise in prices, duties and quotas may be gradually removed. 
This step would have the advantage of retarding the rise in prices and at 
the same time of stimulating growth and efficiency. It would stiffen the 
resistance of employers to wage increases, and at the same time it would 

179 



stimulate the search for cost-saving methods and would encourage the 
shifting of labor and capital from less productive to more productive uses. 
But progress in eliminating duties and quotas will be slow. One fact that 
stands out conspicuously in this discussion of inflation is that ours is a 
producer-dominated economy — the consumer is the forgotten man. We 
have the institutional arrangements that makes gains in productivity pro- 
duce higher wages and higher prices, but no one even speculates about the 
possibility of altering our institutions so that gains in productivity will 
produce lower prices. This absence of concern for the consumer is under- 
standable because the consumer does not demand lower prices. 

Would the several steps that I have suggested check the tendency for 
unions to push up wages faster than the rise in output per man-hour? I do 
not know. Other influences are growing in importance and combine with 
the trade unions to produce rising prices. For example, we stand only on 
the threshold of the age of science. Nine-tenths of all the scientists who 
have ever lived are said to be alive today. Science is likely to stimulate the 
expansion of credit by discovering investment opportunities faster than the 
community generates investment-seeking funds. Furthermore, science will 
create large profits in various parts of the economy and these profits will 
stimulate stiff wage demands by workers throughout industry. The most 
profitable firms will choose to concede much of what the unions ask. 
Hence, wages will continue to outrun output per man-hour. 

Let us try to recover common sense and perspective in dealing with our 
problems. Speaking before the Associated Press recently Secretary 
Robert B. Anderson said, "This country cannot have an enduring bright 
economic future with inflation." That narrow and pessimistic view of our 
economic prospects is unrealistic. No one can be sure whether or not we 
are going to have inflation. But with science and technology rapidly ex- 
pending and with the art of business management rapidly developing, it is 
clear that substantial gains in productivity are ahead. There is a strong 
likelihood that technological progress in conjunction with our strong trade 
unions will be a cause of inflation. But regardless of whether we continue 
to have inflation or not the economic future of America is bright. 



180 



Final report of 

the Fifteenth American Assembly 



At the close of their discussions, the participants in the Fifteenth American 
Assembly, at Arden House, Harriman, New York, April 30 - May 3, 1959, on 
WAGES, PRICES, PROFITS AND PRODUCTIVITY, reviewed as a group 
the following statement. Although there was general agreement on the Final 
Report, it is not the practice of The American Assembly for participants to 
affix their signatures, and it should not be assumed that every participant neces- 
sarily subscribes to every recommendation included in the statement. 



We have examined four main aspects of wage-price-profits-productivity 
relationships: the measures and their limitations, factors in the postwar 
inflation, the compatibility of national economic goals, and policy proposals. 

I. The measures and their limitations 

Informed public discussion of wages, prices, profits and productivity 
depends upon a better understanding of the measures used and their limi- 
tations. There are many types of measures; which should be used in any 
specific case depends upon the purposes of the user. 

Limitations of existing measures were discussed. Wage data, for ex- 
ample, often relate only to production workers. There should also be data 
for non-production workers, further data on supplements to wages and 

181 



salaries ("fringe benefits"), and better data for non-manufacturing indus- 
tries. The Consumer Price Index (formerly called the "cost-of-living 
index") overstates the upward movement of prices, since quality improve- 
ments, which are difficult to measure, cannot be taken fully into account. 
There was general agreement that the Consumer Price Index is so often 
used for wage and salary adjustments today that it needs to be continuously 
improved. Among other limitations, productivity measures tend to under- 
state increases, because quality improvements are difficult to calculate. 
Estimates of profit need a better statistical base, and adjustments to reflect 
changes in accounting procedures. The impact of price changes on de- 
preciation allowances and the relationship between profits and investment 
should be studied. 

At best, available statistical measures are rough, and significance should 
not be attached to small or short-period changes. Care should be used in 
selecting the time periods for which comparisons are made. 

There was general agreement that more financial support is needed for 
governmental and private statistical agencies both for collection of data 
and, even more, for their analysis. It was urged that funds be appropriated 
for a survey of consumer expenditures to provide new weights for the 
Consumer Price Index. Funds should be made available for collection 
and more prompt publication of better data on all resources used in the 
economy, including labor and capital by type and amount actually used. 
Also, funds should be made available for needed knowledge of the com- 
position of unemployment, and compensation of nonproduction workers 
and employees outside of manufacturing. 

II. Factors in the postwar inflation 

Money wages and retail prices have not risen as much in the United 
States as in most other industrial countries since 1948. The base year 
partially explains this difference, but if a more recent base year were 
chosen, our record would still be good by comparison. 

There was considerable discussion of the relative importance of the up- 
ward pull of demand on prices as compared to the upward push of costs 
since the end of the war. It was generally agreed that during the immediate 
postwar period (1945-48) price movements were dominated by demand 
factors — the pent-up demand carried over from the war, and the relatively 
easy money policies. After the outbreak of the Korean War in 1950, 
demand forces were again dominant for a brief period. 

There was general agreement that the 1953-57 period had somewhat 
different characteristics: monetary restraint was greater, demand forces 
less buoyant, output growth less rapid, price rises somewhat smaller, and 
profit margins lower. Wages continued to increase. The investment boom 

182 



in 1955-57 created a demand-pull for many producers' goods, and the 
shifts in consumer demand toward services (for example, medical services) 
contributed to the rising price of services, so that cost and demand factors 
were both operating. However, the continued advance of prices and wages 
in the late 1957-58 period could hardly be explained by general excess 
demand. 

There were differences of opinion regarding the relative significance of 
these factors. Some participants placed primary stress on negotiated wage 
increases, especially in those industries where wages seem to have been 
"pushing against an open door," and pointed out that they tend to be 
transmitted in wage increases throughout the labor market and in price 
increases throughout the economy. Others called attention to the price 
policies of firms, particularly in highly concentrated industries, the effects 
of which are also transmitted throughout the economy. The inter-rela- 
tionship of market structure, wage policies, price policies and profits is seen 
clearly in the contrasting experience of the steel and clothing industries, for 
example. Since these forces often operate concurrently and are frequently 
interdependent, many participants felt it may be misleading to blame any 
particular factor or group for increases in labor costs and prices. 

III. The compatibility of national economic goals 

There was considerable discussion of national economic goals and the 
extent to which they are compatible. These include a sustained and strong 
rate of economic growth, full employment, reasonable price stability, na- 
tional security and industrial peace within the framework of our free 
institutions. There was general agreement that the achievement of some 
of these goals may involve the partial sacrifice of others. 

Many felt that the average level of unemployment during 1953-58 of 
4.7 per cent (approximately three million unemployed) was too high. 
Reliance on large-scale unemployment to achieve price stability is intoler- 
able in our present society. If the cost of price stability is a high level of 
unemployment (which may not be the case) many would prefer a mild 
increase in prices. 

There was consensus that the rate of growth in the American economy 
should be increased, even if this interferes with the full attainment of other 
economic goals. Many believed that this would mean a 4 or 5 per cent 
rate instead of the recent lower rate. They shared the view that a lower 
rate of economic growth imposed by an unduly restrictive monetary policy, 
the principal aim of which is price stability, impairs our international posi- 
tion, makes financing much-needed public improvements difficult, involves 
loss of production, increases unemployment, wastes skills, bears unevenly 
upon certain sectors of the economy, and imperils marginal firms. Some, 

183 



however, dissented, holding that a higher rate of growth requires price 
stability, accompanied if necessary by a restrictive monetary policy. 

Some thought that preoccupation with inflation impaired the attainment 
of other economic goals. Many believed that the Consumer Price Index is 
not likely to increase very much in the next several years. A minority felt 
that we will have more inflation unless vigorous action is taken to arrest 
the cost-push as well as to restrain demand. This difference was reflected 
in the discussion of policy choices. 



IV. Policy proposals 

1. We should seek maximum sustained economic growth as a primary 
national objective because it is essential to the improvement of living 
standards and national security. In the pursuit of economic growth, how- 
ever, we should seek to maintain a reasonably stable price level and to hold 
unemployment to the minimum necessary to preserve flexibility in the 
economy. Although the United States has done reasonably well since 1951 
in maintaining price stability, we should develop a higher utilization of 
the labor force and achieve faster economic growth through increased 
productivity. 

2. It is urgent that management, labor and government take measures 
to accelerate the rate of increase in productivity. Among the steps that can 
be taken are these: increasing expenditures on research and development; 
improving the quality of education; training more scientific, technical and 
managerial personnel; upgrading the skills of the labor force; eliminating 
restrictive practices by labor and management; and stimulating the ideas 
and energies of our work force. 

3. Measures should be taken to increase worker mobility and to im- 
prove the operation of the labor market. Such measures might include 
severance and relocation allowances, retraining programs, employment of 
qualified older workers, transferability of pension rights, the extension of 
seniority units, provision of part-time jobs, and diversification of economic 
activity in distressed areas. 

4. Although reliance on unemployment is not an acceptable method of 
controlling inflation, it is recognized that a dynamic economy brings about 
fluctuations in employment in particular occupations, industries and areas. 

5. Radical change in our present wage-setting machinery would not be 
a fruitful approach to the inflation problem. Foreign experience does not 
suggest the desirability for the United States of economy-wide bargaining 
as an alternative to our more decentralized system. Breaking up national 
unions into smaller units would also not be an effective way of seeking 
greater wage and price stability. 

184 



6. The average increase in output per man-hour in the economy as a 
whole is a relevant consideration in wage negotiations, but in view of the 
numerous other factors to be weighed and reconciled, it cannot be the 
only criterion of a proper adjustment. 

7. Appropriate monetary and fiscal policies are necessary to deal with 
inflation arising from excessive demand. However, sole reliance on these 
measures is inadequate for dealing with price increases in the face of 
excess capacity and unemployment. 

8. Despite the seriousness of our present international position, direct 
general government controls of prices and wages are incompatible with the 
institutions of a free economy. 

9. As one means of achieving price stability, the government should 
work toward a freer market in agricultural products, reflecting advances in 
productivity. Special measures should be provided to ameliorate economic 
hardships upon individuals. 

10. Price stability should be encouraged by reducing tariffs and other 
impediments to international trade. Measures to provide temporary relief 
for areas adversely affected by tariff policy may be appropriate to facilitate 
readjustments. 

11. Effective competition will help bring down prices raised by mo- 
nopoly power. We favor vigorous enforcement of the anti-trust laws. Acts 
in restraint of trade in the market for goods and services, now illegal when 
committed by business firms or by firms and unions jointly, should also 
be illegal when undertaken by unions alone. 

12. Pleas by government officials for restraint and responsibility in 
wage and price decisions are likely to have little effect on the course of 
wage and price movements. 

Some participants thought that industries of importance as pattern setters 
should be subject to impartial fact-finding review of their wage bargains 
and price decisions to acquaint the public with their consequences. How- 
ever, very few supported a procedure delaying the putting into effect of 
wage and price increases in key industries pending public investigation. 

13. An annual conference of labor, management and government repre- 
sentatives should be convened, shortly after the presentation of the Presi- 
dent's Economic Report, to discuss wages, prices, profits and productivity 
as related to national economic goals. These sessions should not be 
concerned with any particular contract negotiations. The objective is to 
reduce the diversity of views about the short-term and long-term economic 
outlook and to discuss appropriate private and public policies to achieve 
growth and stability. 



185 



L 



Participants in the 

Fifteenth American Assembly 



Hugo A. Anderson 

First National Bank of Chicago 

Harry Robert Bartell 
Harriman Scholar 
Columbia University 

Atherton Bean 

President 

International Milling Company 

Minneapolis 

Ralph H. Bergmann 

United Rubber, Cork, Linoleum and 

Plastic Workers of America 
Akron 

Richard W. Bolling 
United States Congressman from 
Missouri 

William G. Bowen 
Industrial Relations Section 
Princeton University 



Courtney C. Brown 

Dean 

Graduate School of Business 

Columbia University 

Thomas H. Carroll 
Vice President 
The Ford Foundation 
New York 

Ewan Clague 

United States Commissioner of 

Labor Statistics 
Washington, D. C. 

Joseph S. Clark 
United States Senator from 
Pennsylvania 

John Cowles 

President 

Minneapolis Star & Tribune 

Minnesota 



187 



Catherine B. Cleary 

Vice President 

First Wisconsin Trust Company 

Milwaukee 

Gerhard Colm 

National Planning Association 

Washington, D. C. 

John T. Dunlop 
Professor of Economics 
Harvard University 

Julian B. Feibelman 
Rabbi 

Temple Sinai 
New Orleans 

Philip J. Fitzgerald 
Dean Witter & Company 
San Francisco 

William C. Foster 

Vice President 

Olin Mathieson Chemical 

Corporation 
Washington, D. C. 

Henry H. Fowler 

Fowler, Leva, Hawes & Symington 

Washington, D. C. 

Joseph Friedman 
Harriman Scholar 
Columbia University 

Edwin B. George 
Dun and Bradstreet, Inc. 
New York 

Ben S. Gilmer 
President 

Southern Bell Telephone & Tele- 
graph Company 
Atlanta 



W. Averell Harriman 
New York 

Leland Hazard 

Professor of Industrial Administra- 
tion and Law 
Carnegie Institute of Technology 

and 
Director-Consultant 
Pittsburgh Plate Glass Company 

Benjamin Higgins 
Professor of Economics 
University of Texas 

George H. Hildebrand 
Professor of Economics 
University of California 
Los Angeles 

John M. Houchin 

Vice President 

Phillips Petroleum Company 

Bartlesville, Oklahoma 

Norris O. Johnson 
Vice President 
First National City Bank of 
New York 

J. M. Kaplan 
New York 

Everett M. Kassalow 
Industrial Union Department 
AFL-CIO 
Washington, D. C. 

Lester S. Kellogg 
Deere & Company 
Moline, Illinois 

John W. Kendrick 

Associate Professor of Economics 

George Washington University 



188 



Clark Kerr 

President 

University of California 

Berkeley 

Robert L. Kirkpatrick 
Kirkpatrick, Pomeroy, Lockhart 

& Johnson 
Pittsburgh 



Charles A. Myers 
Director 

Industrial Relations Section 
Massachusetts Institute 
of Technology 

Robert R. Nathan 

Robert R. Nathan Associates, Inc. 

Washington, D. C. 



Allan B. Kline 
Western Springs, Illinois 

James N. Land 
Senior Vice President 
Mellon National Bank & Trust 

Company 
Pittsburgh 

T. A. Lane 

Major-General, U. S. A. 
Commanding General 
Fort Leonard Wood 

Isador Lubin 

Franklin D. Roosevelt Foundation 

New York 

Daniel May 

Chairman 

May Hosiery Mills 

Nashville 



Fred M. Nelson 

Chairman 

Texas Gulf Sulphur Company 

Houston 

Gustav Peck 

Legislative Reference Service 
Library of Congress 
Washington, D. C. 

Howard C. Petersen 
President 
Fidelity-Philadelphia Trust 

Company 
Philadelphia 

Sylvia F. Porter 
The New York Post 

John Post 

Continental Oil Company 

Houston 



John G. McLean 
Vice President 
Continental Oil Company 
Houston 

William B. Miller 
Executive Secretary 
Town Hall 
Los Angeles 



Jacob S. Potofsky 

President 

Amalgamated Clothing Workers of 

America 
New York 

Albert Rees 

Associate Professor of Economics 

The University of Chicago 



189 



Henry S. Reuss 

United States Congressman 
from Wisconsin 

Benjamin C. Roberts 
Professor of Economics 
London School of Economics 

Arthur M. Ross 

Director 

Institute of Industrial Relations 

University of California 

Berkeley 

Stanley H. Ruttenberg 
Director of Research 
AFL-CIO 
Washington, D. C. 

Charles B. Shuman 

President 

American Farm Bureau Federation 

Chicago 



S. Abbot Smith 

President 

Thomas Strahan Company 

Chelsea, Massachusetts 

Herbert Stein 
Committee for Economic 

Development 
Washington, D. C. 

Nat Weinberg 
United Auto Workers 
Detroit 

John L. Weller 
President 

Seatrain Lines, Inc. 
New York 

The Rev. Hugh C. White, Jr. 
Detroit Industrial Mission 
Michigan 

David J. Winton 
Minneapolis 



Leonard Silk 
Economics Editor 
Business Week 
New York 

Sumner H. Slichter 
Professor of Economics 
Harvard University 



Emmett Wallace 
Harriman Scholar 
Columbia University 

Allen Wallis 
Executive Vice Chairman 
Cabinet Committee on Price Stabil- 
ity for Economic Growth 
The White House 



190 



The American Assembly 



Trustees 



Courtney C. Brown ex officio 


New York 


William Benton 


Connecticut 


Harry A. Bullis 


Minnesota 


John Cowles 


Minnesota 


Thomas K. Finletter 


New York 


W. Averell Harriman 


New York 


Oveta Gulp Hobby 


Texas 


George M. Humphrey 


Ohio 


W. Alton Jones 


New York 


Grayson Kirk ex officio 


New York 


Allan B. Kline 


Illinois 


Harlan Logan 


Connecticut 


Leonard F. McCollum 


Texas 


Don G. Mitchell 


New Jersey 


Reuben B. Robertson, Jr. 


Ohio 


Robert W. Woodruff 


Georgia 


Henry M. Wriston ex officio 


Rhode Island 



Officers 

Henry M. Wriston President 

Clifford C. Nelson Executive Vice President 

Peter C. Grenquist Secretary-Treasurer 

191 



Lhe American Assembly is a program of conferences which bring together 
business, labor, farm groups, the professions, political parties, government 
and the academic community. These meetings develop recommendations 
on issues of national concern. The American Assembly is a non-partisan 
public service designed to throw light on problems confronting citizens of 
the United States. 

Dwight D. Eisenhower, as president of Columbia University, established 
the Assembly in 1950. It is incorporated as an educational institution by 
the Regents of the University of the State of New York. 

Subjects for American Assembly discussion are screened for timeliness 
and importance by a Faculty Advisory Committee of Columbia University 
and considered by the Assembly's Board of Trustees. Leading authorities 
are retained to write background papers presenting the major issues in 
each problem. 

The American Assembly discussions at Arden House, Harriman, New 
York, are based on this background material. The recommendations of the 
Assembly participants and the background papers are published in a Final 
Edition which is distributed to libraries, colleges, government, adult educa- 
tion groups, and organizations of industry, labor and the community for 
further study and discussion. 

National meetings at Arden House have been held on the following 
subjects: 

1951 UNITED STATES — WESTERN EUROPE RELATIONSHIPS 

1952 inflation (two Assemblies) 

1953 ECONOMIC SECURITY FOR AMERICANS 

1954 THE UNITED STATES STAKE IN THE UNITED NATIONS 
THE FEDERAL GOVERNMENT SERVICE 

1955 UNITED STATES AGRICULTURE 

the forty-eight states (State Government) 

1956 THE REPRESENTATION OF THE UNITED STATES ABROAD 
THE UNITED STATES AND THE FAR EAST 

1957 international stability and progress (Foreign Aid) 

ATOMS FOR POWER 

1958 THE UNITED STATES AND AFRICA 
UNITED STATES MONETARY POLICY 

1959 WAGES, PRICES, PROFITS AND PRODUCTIVITY 

192 



Follow-up sessions on Assembly subjects are also held on the regional, 
state and local level. Institutions which, with Assembly staff cooperation, 
have sponsored, or are scheduled to sponsor, these meetings are: 

Stanford University 

University of California (Berkeley) 

University of California (Los Angeles) 

University of Wyoming 

University of Denver 

University of New Mexico 

University of Oklahoma 

Dallas World Affairs Council 

The Rice Institute 

Tulane University 

Southwestern at Memphis 

Duke University 

University of Florida 

Emory University 

University of Illinois 

Minnesota World Affairs Center 

University of Washington 

Cleveland Council on World Affairs 

University of Missouri 

Washington University 

Drake University 

Indiana University 

University of Vermont 

Tufts University 

Foreign Policy Association of Pittsburgh 

Southern Methodist University 

University of Texas 

Town Hall of Los Angeles 

North Central Association 

United States Air Force Academy 

University of Arkansas 

International Relations Council of Kansas City 

Vanderbilt University 

Michigan State University 

The University of Puerto Rico 

Lawrence College 



193 



^ue 


Date 
Returned 


Due 
Due 


Returned 


tmWn 


/ 






! Jfi 1 '60 






REUASEt 


» *wBJ 






RESER 


VE 






/ 


\ 






/ 


\ 








.. V 






rtpi<>70 RE 








FEB 


2 5 1970 














DEC 2Q TW 


fit i SIS' 




















































/ 








/ 




»^ 




1 
















J 




£ 


r 


1 








/ 























33?, 9-A. 



Wages, prices, profits, and pr main 
338.082A512wC2 



3 lEbE 032fl0 552^4 



KEEP CARD IN POCKET 





Late 


Due 




DUE 


RETURNED 


DUE ^^LjETURNED 




J 


4 JT 


A 




-^4 


f € 


1 


t 


^Ff\ 


f * 


J 


iJjL 






/ 




/Jm 


ff 


1/ 








1 


l 'rW \ 






[• j 


W 1 






■ -£ 












J L 




' ' ^K "'»• 


* 


\* * 




pm 


< 






: 4>* 


j^^p 






^^^H 














^ 










.. _ . _ ,. , SJK